[llvm] 83ccce6 - [X86] Add some add reduction tests when we're extending from a smaller type
Simon Pilgrim via llvm-commits
llvm-commits at lists.llvm.org
Sun Feb 13 05:27:19 PST 2022
Author: Simon Pilgrim
Date: 2022-02-13T13:27:00Z
New Revision: 83ccce6cedbd32f13f73e9805ccbbd1e97dadbfa
URL: https://github.com/llvm/llvm-project/commit/83ccce6cedbd32f13f73e9805ccbbd1e97dadbfa
DIFF: https://github.com/llvm/llvm-project/commit/83ccce6cedbd32f13f73e9805ccbbd1e97dadbfa.diff
LOG: [X86] Add some add reduction tests when we're extending from a smaller type
This often occurs when the accumulator is larger than the source vector element, such as some of the tests mentioned in Issue #42019
Added:
llvm/test/CodeGen/X86/vector-reduce-add-sext.ll
llvm/test/CodeGen/X86/vector-reduce-add-zext.ll
Modified:
Removed:
################################################################################
diff --git a/llvm/test/CodeGen/X86/vector-reduce-add-sext.ll b/llvm/test/CodeGen/X86/vector-reduce-add-sext.ll
new file mode 100644
index 0000000000000..3c51fda1531d1
--- /dev/null
+++ b/llvm/test/CodeGen/X86/vector-reduce-add-sext.ll
@@ -0,0 +1,1518 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefixes=SSE2
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse4.1 | FileCheck %s --check-prefixes=SSE41
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefixes=AVX,AVX1,AVX1-SLOW
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx,+fast-hops | FileCheck %s --check-prefixes=AVX,AVX1,AVX1-FAST
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefixes=AVX,AVX2
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512bw | FileCheck %s --check-prefix=AVX512
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512bw,+avx512vl | FileCheck %s --check-prefix=AVX512
+
+;
+; vXi64
+;
+
+define i64 @test_v2i64_v2i32(<2 x i32> %a0) {
+; SSE2-LABEL: test_v2i64_v2i32:
+; SSE2: # %bb.0:
+; SSE2-NEXT: pxor %xmm1, %xmm1
+; SSE2-NEXT: pcmpgtd %xmm0, %xmm1
+; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddq %xmm0, %xmm1
+; SSE2-NEXT: movq %xmm1, %rax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v2i64_v2i32:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovsxdq %xmm0, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddq %xmm0, %xmm1
+; SSE41-NEXT: movq %xmm1, %rax
+; SSE41-NEXT: retq
+;
+; AVX-LABEL: test_v2i64_v2i32:
+; AVX: # %bb.0:
+; AVX-NEXT: vpmovsxdq %xmm0, %xmm0
+; AVX-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX-NEXT: vmovq %xmm0, %rax
+; AVX-NEXT: retq
+;
+; AVX512-LABEL: test_v2i64_v2i32:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovsxdq %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovq %xmm0, %rax
+; AVX512-NEXT: retq
+ %1 = sext <2 x i32> %a0 to <2 x i64>
+ %2 = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %1)
+ ret i64 %2
+}
+
+define i64 @test_v4i64_v4i16(<4 x i16> %a0) {
+; SSE2-LABEL: test_v4i64_v4i16:
+; SSE2: # %bb.0:
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
+; SSE2-NEXT: psrad $16, %xmm0
+; SSE2-NEXT: pxor %xmm1, %xmm1
+; SSE2-NEXT: pcmpgtd %xmm0, %xmm1
+; SSE2-NEXT: movdqa %xmm0, %xmm2
+; SSE2-NEXT: punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm1[2],xmm2[3],xmm1[3]
+; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE2-NEXT: paddq %xmm2, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddq %xmm0, %xmm1
+; SSE2-NEXT: movq %xmm1, %rax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v4i64_v4i16:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovsxwq %xmm0, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
+; SSE41-NEXT: pmovsxwq %xmm0, %xmm0
+; SSE41-NEXT: paddq %xmm1, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddq %xmm0, %xmm1
+; SSE41-NEXT: movq %xmm1, %rax
+; SSE41-NEXT: retq
+;
+; AVX1-LABEL: test_v4i64_v4i16:
+; AVX1: # %bb.0:
+; AVX1-NEXT: vpmovsxwq %xmm0, %xmm1
+; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
+; AVX1-NEXT: vpmovsxwq %xmm0, %xmm0
+; AVX1-NEXT: vpaddq %xmm0, %xmm1, %xmm0
+; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vmovq %xmm0, %rax
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: test_v4i64_v4i16:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovsxwq %xmm0, %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovq %xmm0, %rax
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v4i64_v4i16:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovsxwq %xmm0, %ymm0
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovq %xmm0, %rax
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %1 = sext <4 x i16> %a0 to <4 x i64>
+ %2 = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %1)
+ ret i64 %2
+}
+
+define i64 @test_v8i64_v8i8(<8 x i8> %a0) {
+; SSE2-LABEL: test_v8i64_v8i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm1 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
+; SSE2-NEXT: psrad $24, %xmm1
+; SSE2-NEXT: pxor %xmm2, %xmm2
+; SSE2-NEXT: pxor %xmm3, %xmm3
+; SSE2-NEXT: pcmpgtd %xmm1, %xmm3
+; SSE2-NEXT: movdqa %xmm1, %xmm4
+; SSE2-NEXT: punpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
+; SSE2-NEXT: psrad $24, %xmm0
+; SSE2-NEXT: pcmpgtd %xmm0, %xmm2
+; SSE2-NEXT: movdqa %xmm0, %xmm5
+; SSE2-NEXT: punpckldq {{.*#+}} xmm5 = xmm5[0],xmm2[0],xmm5[1],xmm2[1]
+; SSE2-NEXT: punpckhdq {{.*#+}} xmm1 = xmm1[2],xmm3[2],xmm1[3],xmm3[3]
+; SSE2-NEXT: punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
+; SSE2-NEXT: paddq %xmm1, %xmm0
+; SSE2-NEXT: paddq %xmm4, %xmm0
+; SSE2-NEXT: paddq %xmm5, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddq %xmm0, %xmm1
+; SSE2-NEXT: movq %xmm1, %rax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v8i64_v8i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovsxbq %xmm0, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm0[1,1,1,1]
+; SSE41-NEXT: pmovsxbq %xmm2, %xmm2
+; SSE41-NEXT: movdqa %xmm0, %xmm3
+; SSE41-NEXT: psrlq $48, %xmm3
+; SSE41-NEXT: pmovsxbq %xmm3, %xmm3
+; SSE41-NEXT: psrld $16, %xmm0
+; SSE41-NEXT: pmovsxbq %xmm0, %xmm0
+; SSE41-NEXT: paddq %xmm3, %xmm0
+; SSE41-NEXT: paddq %xmm2, %xmm0
+; SSE41-NEXT: paddq %xmm1, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddq %xmm0, %xmm1
+; SSE41-NEXT: movq %xmm1, %rax
+; SSE41-NEXT: retq
+;
+; AVX1-LABEL: test_v8i64_v8i8:
+; AVX1: # %bb.0:
+; AVX1-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX1-NEXT: vpmovsxbq %xmm1, %xmm1
+; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm0[1,1,1,1]
+; AVX1-NEXT: vpsrld $16, %xmm2, %xmm3
+; AVX1-NEXT: vpmovsxbq %xmm3, %xmm3
+; AVX1-NEXT: vpaddq %xmm3, %xmm1, %xmm1
+; AVX1-NEXT: vpmovsxbq %xmm0, %xmm0
+; AVX1-NEXT: vpmovsxbq %xmm2, %xmm2
+; AVX1-NEXT: vpaddq %xmm1, %xmm2, %xmm1
+; AVX1-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vmovq %xmm0, %rax
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: test_v8i64_v8i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovsxbq %xmm0, %ymm1
+; AVX2-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpmovsxbq %xmm0, %ymm0
+; AVX2-NEXT: vpaddq %ymm0, %ymm1, %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovq %xmm0, %rax
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v8i64_v8i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovsxbq %xmm0, %zmm0
+; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm1
+; AVX512-NEXT: vpaddq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovq %xmm0, %rax
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %1 = sext <8 x i8> %a0 to <8 x i64>
+ %2 = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %1)
+ ret i64 %2
+}
+
+define i64 @test_v16i64_v16i8(<16 x i8> %a0) {
+; SSE2-LABEL: test_v16i64_v16i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm8 = xmm8[8],xmm0[8],xmm8[9],xmm0[9],xmm8[10],xmm0[10],xmm8[11],xmm0[11],xmm8[12],xmm0[12],xmm8[13],xmm0[13],xmm8[14],xmm0[14],xmm8[15],xmm0[15]
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm8[0],xmm1[1],xmm8[1],xmm1[2],xmm8[2],xmm1[3],xmm8[3]
+; SSE2-NEXT: psrad $24, %xmm1
+; SSE2-NEXT: pxor %xmm10, %xmm10
+; SSE2-NEXT: pxor %xmm12, %xmm12
+; SSE2-NEXT: pcmpgtd %xmm1, %xmm12
+; SSE2-NEXT: movdqa %xmm1, %xmm11
+; SSE2-NEXT: punpckhdq {{.*#+}} xmm11 = xmm11[2],xmm12[2],xmm11[3],xmm12[3]
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm9 = xmm9[0],xmm0[0],xmm9[1],xmm0[1],xmm9[2],xmm0[2],xmm9[3],xmm0[3],xmm9[4],xmm0[4],xmm9[5],xmm0[5],xmm9[6],xmm0[6],xmm9[7],xmm0[7]
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm9[0],xmm0[1],xmm9[1],xmm0[2],xmm9[2],xmm0[3],xmm9[3]
+; SSE2-NEXT: psrad $24, %xmm0
+; SSE2-NEXT: pxor %xmm7, %xmm7
+; SSE2-NEXT: pcmpgtd %xmm0, %xmm7
+; SSE2-NEXT: movdqa %xmm0, %xmm13
+; SSE2-NEXT: punpckhdq {{.*#+}} xmm13 = xmm13[2],xmm7[2],xmm13[3],xmm7[3]
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm6 = xmm6[4],xmm8[4],xmm6[5],xmm8[5],xmm6[6],xmm8[6],xmm6[7],xmm8[7]
+; SSE2-NEXT: psrad $24, %xmm6
+; SSE2-NEXT: pxor %xmm2, %xmm2
+; SSE2-NEXT: pcmpgtd %xmm6, %xmm2
+; SSE2-NEXT: movdqa %xmm6, %xmm5
+; SSE2-NEXT: punpckhdq {{.*#+}} xmm5 = xmm5[2],xmm2[2],xmm5[3],xmm2[3]
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm9[4],xmm3[5],xmm9[5],xmm3[6],xmm9[6],xmm3[7],xmm9[7]
+; SSE2-NEXT: psrad $24, %xmm3
+; SSE2-NEXT: pcmpgtd %xmm3, %xmm10
+; SSE2-NEXT: movdqa %xmm3, %xmm4
+; SSE2-NEXT: punpckhdq {{.*#+}} xmm4 = xmm4[2],xmm10[2],xmm4[3],xmm10[3]
+; SSE2-NEXT: paddq %xmm5, %xmm4
+; SSE2-NEXT: paddq %xmm11, %xmm4
+; SSE2-NEXT: paddq %xmm13, %xmm4
+; SSE2-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm12[0],xmm1[1],xmm12[1]
+; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm7[0],xmm0[1],xmm7[1]
+; SSE2-NEXT: punpckldq {{.*#+}} xmm6 = xmm6[0],xmm2[0],xmm6[1],xmm2[1]
+; SSE2-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm10[0],xmm3[1],xmm10[1]
+; SSE2-NEXT: paddq %xmm6, %xmm3
+; SSE2-NEXT: paddq %xmm1, %xmm3
+; SSE2-NEXT: paddq %xmm4, %xmm3
+; SSE2-NEXT: paddq %xmm0, %xmm3
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm3[2,3,2,3]
+; SSE2-NEXT: paddq %xmm3, %xmm0
+; SSE2-NEXT: movq %xmm0, %rax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v16i64_v16i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: movdqa %xmm0, %xmm1
+; SSE41-NEXT: movdqa %xmm0, %xmm2
+; SSE41-NEXT: psrldq {{.*#+}} xmm2 = xmm2[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; SSE41-NEXT: pmovsxbq %xmm2, %xmm2
+; SSE41-NEXT: movdqa %xmm0, %xmm3
+; SSE41-NEXT: psrlq $48, %xmm3
+; SSE41-NEXT: pmovsxbq %xmm3, %xmm3
+; SSE41-NEXT: paddq %xmm2, %xmm3
+; SSE41-NEXT: pmovsxbq %xmm0, %xmm2
+; SSE41-NEXT: pshufd {{.*#+}} xmm4 = xmm0[2,3,2,3]
+; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm0[3,3,3,3]
+; SSE41-NEXT: pshufd {{.*#+}} xmm6 = xmm0[1,1,1,1]
+; SSE41-NEXT: psrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; SSE41-NEXT: pmovsxbq %xmm0, %xmm0
+; SSE41-NEXT: paddq %xmm0, %xmm3
+; SSE41-NEXT: psrld $16, %xmm1
+; SSE41-NEXT: pmovsxbq %xmm1, %xmm0
+; SSE41-NEXT: paddq %xmm0, %xmm3
+; SSE41-NEXT: pmovsxbq %xmm4, %xmm0
+; SSE41-NEXT: pmovsxbq %xmm5, %xmm1
+; SSE41-NEXT: pmovsxbq %xmm6, %xmm4
+; SSE41-NEXT: paddq %xmm1, %xmm4
+; SSE41-NEXT: paddq %xmm0, %xmm4
+; SSE41-NEXT: paddq %xmm3, %xmm4
+; SSE41-NEXT: paddq %xmm2, %xmm4
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm4[2,3,2,3]
+; SSE41-NEXT: paddq %xmm4, %xmm0
+; SSE41-NEXT: movq %xmm0, %rax
+; SSE41-NEXT: retq
+;
+; AVX1-LABEL: test_v16i64_v16i8:
+; AVX1: # %bb.0:
+; AVX1-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX1-NEXT: vpmovsxbq %xmm1, %xmm1
+; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm0[2,3,2,3]
+; AVX1-NEXT: vpmovsxbw %xmm2, %xmm3
+; AVX1-NEXT: vpshufd {{.*#+}} xmm4 = xmm3[1,1,1,1]
+; AVX1-NEXT: vpmovsxwq %xmm4, %xmm4
+; AVX1-NEXT: vpmovsxbw %xmm0, %xmm5
+; AVX1-NEXT: vpshufd {{.*#+}} xmm6 = xmm5[3,3,3,3]
+; AVX1-NEXT: vpmovsxwq %xmm6, %xmm6
+; AVX1-NEXT: vpshufd {{.*#+}} xmm7 = xmm3[3,3,3,3]
+; AVX1-NEXT: vpmovsxwq %xmm7, %xmm7
+; AVX1-NEXT: vpaddq %xmm7, %xmm6, %xmm6
+; AVX1-NEXT: vpaddq %xmm6, %xmm4, %xmm4
+; AVX1-NEXT: vpaddq %xmm4, %xmm1, %xmm1
+; AVX1-NEXT: vpshufd {{.*#+}} xmm4 = xmm5[2,3,2,3]
+; AVX1-NEXT: vpmovsxwq %xmm4, %xmm4
+; AVX1-NEXT: vpshufd {{.*#+}} xmm3 = xmm3[2,3,2,3]
+; AVX1-NEXT: vpmovsxwq %xmm3, %xmm3
+; AVX1-NEXT: vpaddq %xmm3, %xmm4, %xmm3
+; AVX1-NEXT: vpmovsxbq %xmm0, %xmm0
+; AVX1-NEXT: vpmovsxbq %xmm2, %xmm2
+; AVX1-NEXT: vpaddq %xmm3, %xmm2, %xmm2
+; AVX1-NEXT: vpaddq %xmm1, %xmm2, %xmm1
+; AVX1-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vmovq %xmm0, %rax
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: test_v16i64_v16i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovsxbw %xmm0, %ymm1
+; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm2
+; AVX2-NEXT: vpshufd {{.*#+}} xmm3 = xmm2[2,3,2,3]
+; AVX2-NEXT: vpmovsxwq %xmm3, %ymm3
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[2,3,2,3]
+; AVX2-NEXT: vpmovsxwq %xmm1, %ymm1
+; AVX2-NEXT: vpaddq %ymm3, %ymm1, %ymm1
+; AVX2-NEXT: vpmovsxbq %xmm0, %ymm0
+; AVX2-NEXT: vpmovsxwq %xmm2, %ymm2
+; AVX2-NEXT: vpaddq %ymm1, %ymm2, %ymm1
+; AVX2-NEXT: vpaddq %ymm1, %ymm0, %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovq %xmm0, %rax
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v16i64_v16i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovsxbw %xmm0, %ymm0
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpmovsxwq %xmm1, %zmm1
+; AVX512-NEXT: vpmovsxwq %xmm0, %zmm0
+; AVX512-NEXT: vpaddq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm1
+; AVX512-NEXT: vpaddq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovq %xmm0, %rax
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %1 = sext <16 x i8> %a0 to <16 x i64>
+ %2 = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %1)
+ ret i64 %2
+}
+
+;
+; vXi32
+;
+
+define i32 @test_v2i32_v2i16(<2 x i16> %a0) {
+; SSE2-LABEL: test_v2i32_v2i16:
+; SSE2: # %bb.0:
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
+; SSE2-NEXT: psrad $16, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; SSE2-NEXT: paddd %xmm0, %xmm1
+; SSE2-NEXT: movd %xmm1, %eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v2i32_v2i16:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovsxwd %xmm0, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; SSE41-NEXT: paddd %xmm0, %xmm1
+; SSE41-NEXT: movd %xmm1, %eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v2i32_v2i16:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vpmovsxwd %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v2i32_v2i16:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vpmovsxwd %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddd %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v2i32_v2i16:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovsxwd %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v2i32_v2i16:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovsxwd %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: retq
+ %1 = sext <2 x i16> %a0 to <2 x i32>
+ %2 = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %1)
+ ret i32 %2
+}
+
+define i32 @test_v4i32(<4 x i8> %a0) {
+; SSE2-LABEL: test_v4i32:
+; SSE2: # %bb.0:
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
+; SSE2-NEXT: psrad $24, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddd %xmm0, %xmm1
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE2-NEXT: paddd %xmm1, %xmm0
+; SSE2-NEXT: movd %xmm0, %eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v4i32:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovsxbd %xmm0, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddd %xmm0, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE41-NEXT: paddd %xmm1, %xmm0
+; SSE41-NEXT: movd %xmm0, %eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v4i32:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vpmovsxbd %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v4i32:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vpmovsxbd %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddd %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddd %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v4i32:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovsxbd %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v4i32:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovsxbd %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: retq
+ %1 = sext <4 x i8> %a0 to <4 x i32>
+ %2 = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %1)
+ ret i32 %2
+}
+
+define i32 @test_v8i32_v8i8(<8 x i8> %a0) {
+; SSE2-LABEL: test_v8i32_v8i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm1 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
+; SSE2-NEXT: psrad $24, %xmm1
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
+; SSE2-NEXT: psrad $24, %xmm0
+; SSE2-NEXT: paddd %xmm1, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddd %xmm0, %xmm1
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE2-NEXT: paddd %xmm1, %xmm0
+; SSE2-NEXT: movd %xmm0, %eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v8i32_v8i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovsxbd %xmm0, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
+; SSE41-NEXT: pmovsxbd %xmm0, %xmm0
+; SSE41-NEXT: paddd %xmm1, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddd %xmm0, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE41-NEXT: paddd %xmm1, %xmm0
+; SSE41-NEXT: movd %xmm0, %eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v8i32_v8i8:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vpmovsxbd %xmm0, %xmm1
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpmovsxbd %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpaddd %xmm0, %xmm1, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v8i32_v8i8:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vpmovsxbd %xmm0, %xmm1
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
+; AVX1-FAST-NEXT: vpmovsxbd %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddd %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddd %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddd %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v8i32_v8i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovsxbd %xmm0, %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v8i32_v8i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovsxbd %xmm0, %ymm0
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %1 = sext <8 x i8> %a0 to <8 x i32>
+ %2 = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %1)
+ ret i32 %2
+}
+
+define i32 @test_v16i32_v16i8(<16 x i8> %a0) {
+; SSE2-LABEL: test_v16i32_v16i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
+; SSE2-NEXT: psrad $24, %xmm2
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
+; SSE2-NEXT: psrad $24, %xmm3
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm1 = xmm1[4,4,5,5,6,6,7,7]
+; SSE2-NEXT: psrad $24, %xmm1
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm0 = xmm0[4,4,5,5,6,6,7,7]
+; SSE2-NEXT: psrad $24, %xmm0
+; SSE2-NEXT: paddd %xmm1, %xmm0
+; SSE2-NEXT: paddd %xmm2, %xmm0
+; SSE2-NEXT: paddd %xmm3, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddd %xmm0, %xmm1
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE2-NEXT: paddd %xmm1, %xmm0
+; SSE2-NEXT: movd %xmm0, %eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v16i32_v16i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovsxbd %xmm0, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm0[2,3,2,3]
+; SSE41-NEXT: pmovsxbd %xmm2, %xmm2
+; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm0[3,3,3,3]
+; SSE41-NEXT: pmovsxbd %xmm3, %xmm3
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
+; SSE41-NEXT: pmovsxbd %xmm0, %xmm0
+; SSE41-NEXT: paddd %xmm3, %xmm0
+; SSE41-NEXT: paddd %xmm2, %xmm0
+; SSE41-NEXT: paddd %xmm1, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddd %xmm0, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE41-NEXT: paddd %xmm1, %xmm0
+; SSE41-NEXT: movd %xmm0, %eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v16i32_v16i8:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vpmovsxbd %xmm0, %xmm1
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm2 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpmovsxbd %xmm2, %xmm2
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm3 = xmm0[3,3,3,3]
+; AVX1-SLOW-NEXT: vpmovsxbd %xmm3, %xmm3
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpmovsxbd %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpaddd %xmm3, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpaddd %xmm0, %xmm2, %xmm0
+; AVX1-SLOW-NEXT: vpaddd %xmm0, %xmm1, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v16i32_v16i8:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vpmovsxbd %xmm0, %xmm1
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm2 = xmm0[2,3,2,3]
+; AVX1-FAST-NEXT: vpmovsxbd %xmm2, %xmm2
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm3 = xmm0[3,3,3,3]
+; AVX1-FAST-NEXT: vpmovsxbd %xmm3, %xmm3
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
+; AVX1-FAST-NEXT: vpmovsxbd %xmm0, %xmm0
+; AVX1-FAST-NEXT: vpaddd %xmm3, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vpaddd %xmm0, %xmm2, %xmm0
+; AVX1-FAST-NEXT: vpaddd %xmm0, %xmm1, %xmm0
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-FAST-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddd %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v16i32_v16i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovsxbd %xmm0, %ymm1
+; AVX2-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpmovsxbd %xmm0, %ymm0
+; AVX2-NEXT: vpaddd %ymm0, %ymm1, %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v16i32_v16i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovsxbd %xmm0, %zmm0
+; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm1
+; AVX512-NEXT: vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %1 = sext <16 x i8> %a0 to <16 x i32>
+ %2 = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %1)
+ ret i32 %2
+}
+
+define i32 @test_v32i32_v32i8(<32 x i8> %a0) {
+; SSE2-LABEL: test_v32i32_v32i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm2[4],xmm3[5],xmm2[5],xmm3[6],xmm2[6],xmm3[7],xmm2[7]
+; SSE2-NEXT: psrad $24, %xmm3
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1],xmm4[2],xmm0[2],xmm4[3],xmm0[3],xmm4[4],xmm0[4],xmm4[5],xmm0[5],xmm4[6],xmm0[6],xmm4[7],xmm0[7]
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm5 = xmm5[4],xmm4[4],xmm5[5],xmm4[5],xmm5[6],xmm4[6],xmm5[7],xmm4[7]
+; SSE2-NEXT: psrad $24, %xmm5
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm1 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm6 = xmm6[4],xmm1[4],xmm6[5],xmm1[5],xmm6[6],xmm1[6],xmm6[7],xmm1[7]
+; SSE2-NEXT: psrad $24, %xmm6
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm0 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm7 = xmm7[4],xmm0[4],xmm7[5],xmm0[5],xmm7[6],xmm0[6],xmm7[7],xmm0[7]
+; SSE2-NEXT: psrad $24, %xmm7
+; SSE2-NEXT: paddd %xmm6, %xmm7
+; SSE2-NEXT: paddd %xmm3, %xmm7
+; SSE2-NEXT: paddd %xmm5, %xmm7
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3]
+; SSE2-NEXT: psrad $24, %xmm2
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3]
+; SSE2-NEXT: psrad $24, %xmm3
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3]
+; SSE2-NEXT: psrad $24, %xmm1
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
+; SSE2-NEXT: psrad $24, %xmm0
+; SSE2-NEXT: paddd %xmm1, %xmm0
+; SSE2-NEXT: paddd %xmm2, %xmm0
+; SSE2-NEXT: paddd %xmm7, %xmm0
+; SSE2-NEXT: paddd %xmm3, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddd %xmm0, %xmm1
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE2-NEXT: paddd %xmm1, %xmm0
+; SSE2-NEXT: movd %xmm0, %eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v32i32_v32i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm1[1,1,1,1]
+; SSE41-NEXT: pmovsxbd %xmm2, %xmm2
+; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm0[1,1,1,1]
+; SSE41-NEXT: pmovsxbd %xmm3, %xmm3
+; SSE41-NEXT: pshufd {{.*#+}} xmm4 = xmm1[3,3,3,3]
+; SSE41-NEXT: pmovsxbd %xmm4, %xmm4
+; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm0[3,3,3,3]
+; SSE41-NEXT: pmovsxbd %xmm5, %xmm5
+; SSE41-NEXT: paddd %xmm4, %xmm5
+; SSE41-NEXT: paddd %xmm2, %xmm5
+; SSE41-NEXT: paddd %xmm3, %xmm5
+; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm1[2,3,2,3]
+; SSE41-NEXT: pmovsxbd %xmm2, %xmm2
+; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm0[2,3,2,3]
+; SSE41-NEXT: pmovsxbd %xmm3, %xmm3
+; SSE41-NEXT: paddd %xmm2, %xmm3
+; SSE41-NEXT: pmovsxbd %xmm1, %xmm1
+; SSE41-NEXT: paddd %xmm3, %xmm1
+; SSE41-NEXT: paddd %xmm5, %xmm1
+; SSE41-NEXT: pmovsxbd %xmm0, %xmm0
+; SSE41-NEXT: paddd %xmm1, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddd %xmm0, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE41-NEXT: paddd %xmm1, %xmm0
+; SSE41-NEXT: movd %xmm0, %eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v32i32_v32i8:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vextractf128 $1, %ymm0, %xmm1
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm2 = xmm1[1,1,1,1]
+; AVX1-SLOW-NEXT: vpmovsxbd %xmm2, %xmm2
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm3 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpmovsxbd %xmm3, %xmm3
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm4 = xmm1[3,3,3,3]
+; AVX1-SLOW-NEXT: vpmovsxbd %xmm4, %xmm4
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm5 = xmm0[3,3,3,3]
+; AVX1-SLOW-NEXT: vpmovsxbd %xmm5, %xmm5
+; AVX1-SLOW-NEXT: vpaddd %xmm4, %xmm5, %xmm4
+; AVX1-SLOW-NEXT: vpaddd %xmm4, %xmm2, %xmm2
+; AVX1-SLOW-NEXT: vpaddd %xmm2, %xmm3, %xmm2
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm3 = xmm1[2,3,2,3]
+; AVX1-SLOW-NEXT: vpmovsxbd %xmm3, %xmm3
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm4 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpmovsxbd %xmm4, %xmm4
+; AVX1-SLOW-NEXT: vpaddd %xmm3, %xmm4, %xmm3
+; AVX1-SLOW-NEXT: vpmovsxbd %xmm1, %xmm1
+; AVX1-SLOW-NEXT: vpaddd %xmm3, %xmm1, %xmm1
+; AVX1-SLOW-NEXT: vpaddd %xmm2, %xmm1, %xmm1
+; AVX1-SLOW-NEXT: vpmovsxbd %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: vzeroupper
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v32i32_v32i8:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vextractf128 $1, %ymm0, %xmm1
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm2 = xmm1[1,1,1,1]
+; AVX1-FAST-NEXT: vpmovsxbd %xmm2, %xmm2
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm3 = xmm0[1,1,1,1]
+; AVX1-FAST-NEXT: vpmovsxbd %xmm3, %xmm3
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm4 = xmm1[3,3,3,3]
+; AVX1-FAST-NEXT: vpmovsxbd %xmm4, %xmm4
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm5 = xmm0[3,3,3,3]
+; AVX1-FAST-NEXT: vpmovsxbd %xmm5, %xmm5
+; AVX1-FAST-NEXT: vpaddd %xmm4, %xmm5, %xmm4
+; AVX1-FAST-NEXT: vpaddd %xmm4, %xmm2, %xmm2
+; AVX1-FAST-NEXT: vpaddd %xmm2, %xmm3, %xmm2
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm3 = xmm1[2,3,2,3]
+; AVX1-FAST-NEXT: vpmovsxbd %xmm3, %xmm3
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm4 = xmm0[2,3,2,3]
+; AVX1-FAST-NEXT: vpmovsxbd %xmm4, %xmm4
+; AVX1-FAST-NEXT: vpaddd %xmm3, %xmm4, %xmm3
+; AVX1-FAST-NEXT: vpmovsxbd %xmm1, %xmm1
+; AVX1-FAST-NEXT: vpaddd %xmm3, %xmm1, %xmm1
+; AVX1-FAST-NEXT: vpaddd %xmm2, %xmm1, %xmm1
+; AVX1-FAST-NEXT: vpmovsxbd %xmm0, %xmm0
+; AVX1-FAST-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-FAST-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddd %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: vzeroupper
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v32i32_v32i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpshufd {{.*#+}} xmm2 = xmm1[2,3,2,3]
+; AVX2-NEXT: vpmovsxbd %xmm2, %ymm2
+; AVX2-NEXT: vpshufd {{.*#+}} xmm3 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpmovsxbd %xmm3, %ymm3
+; AVX2-NEXT: vpaddd %ymm2, %ymm3, %ymm2
+; AVX2-NEXT: vpmovsxbd %xmm1, %ymm1
+; AVX2-NEXT: vpaddd %ymm2, %ymm1, %ymm1
+; AVX2-NEXT: vpmovsxbd %xmm0, %ymm0
+; AVX2-NEXT: vpaddd %ymm1, %ymm0, %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v32i32_v32i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpmovsxbd %xmm1, %zmm1
+; AVX512-NEXT: vpmovsxbd %xmm0, %zmm0
+; AVX512-NEXT: vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm1
+; AVX512-NEXT: vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %1 = sext <32 x i8> %a0 to <32 x i32>
+ %2 = call i32 @llvm.vector.reduce.add.v32i32(<32 x i32> %1)
+ ret i32 %2
+}
+
+;
+; vXi16
+;
+
+define i16 @test_v2i16_v2i8(<2 x i8> %a0) {
+; SSE2-LABEL: test_v2i16_v2i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE2-NEXT: psraw $8, %xmm0
+; SSE2-NEXT: movdqa %xmm0, %xmm1
+; SSE2-NEXT: psrld $16, %xmm1
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: movd %xmm1, %eax
+; SSE2-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v2i16_v2i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovsxbw %xmm0, %xmm0
+; SSE41-NEXT: movdqa %xmm0, %xmm1
+; SSE41-NEXT: psrld $16, %xmm1
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: movd %xmm1, %eax
+; SSE41-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v2i16_v2i8:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vpmovsxbw %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v2i16_v2i8:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vpmovsxbw %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v2i16_v2i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovsxbw %xmm0, %xmm0
+; AVX2-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v2i16_v2i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovsxbw %xmm0, %xmm0
+; AVX512-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX512-NEXT: retq
+ %1 = sext <2 x i8> %a0 to <2 x i16>
+ %2 = call i16 @llvm.vector.reduce.add.v2i16(<2 x i16> %1)
+ ret i16 %2
+}
+
+define i16 @test_v4i16_v4i8(<4 x i8> %a0) {
+; SSE2-LABEL: test_v4i16_v4i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE2-NEXT: psraw $8, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: movdqa %xmm1, %xmm0
+; SSE2-NEXT: psrld $16, %xmm0
+; SSE2-NEXT: paddw %xmm1, %xmm0
+; SSE2-NEXT: movd %xmm0, %eax
+; SSE2-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v4i16_v4i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovsxbw %xmm0, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: movdqa %xmm1, %xmm0
+; SSE41-NEXT: psrld $16, %xmm0
+; SSE41-NEXT: paddw %xmm1, %xmm0
+; SSE41-NEXT: movd %xmm0, %eax
+; SSE41-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v4i16_v4i8:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vpmovsxbw %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v4i16_v4i8:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vpmovsxbw %xmm0, %xmm0
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-FAST-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v4i16_v4i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovsxbw %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v4i16_v4i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovsxbw %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX512-NEXT: retq
+ %1 = sext <4 x i8> %a0 to <4 x i16>
+ %2 = call i16 @llvm.vector.reduce.add.v4i16(<4 x i16> %1)
+ ret i16 %2
+
+}
+
+define i16 @test_v8i16_v8i8(<8 x i8> %a0) {
+; SSE2-LABEL: test_v8i16_v8i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE2-NEXT: psraw $8, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE2-NEXT: paddw %xmm1, %xmm0
+; SSE2-NEXT: movdqa %xmm0, %xmm1
+; SSE2-NEXT: psrld $16, %xmm1
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: movd %xmm1, %eax
+; SSE2-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v8i16_v8i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovsxbw %xmm0, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE41-NEXT: paddw %xmm1, %xmm0
+; SSE41-NEXT: movdqa %xmm0, %xmm1
+; SSE41-NEXT: psrld $16, %xmm1
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: movd %xmm1, %eax
+; SSE41-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v8i16_v8i8:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vpmovsxbw %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v8i16_v8i8:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vpmovsxbw %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v8i16_v8i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovsxbw %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v8i16_v8i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovsxbw %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX512-NEXT: retq
+ %1 = sext <8 x i8> %a0 to <8 x i16>
+ %2 = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %1)
+ ret i16 %2
+}
+
+define i16 @test_v16i16_v16i8(<16 x i8> %a0) {
+; SSE2-LABEL: test_v16i16_v16i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
+; SSE2-NEXT: psraw $8, %xmm1
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE2-NEXT: psraw $8, %xmm0
+; SSE2-NEXT: paddw %xmm1, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE2-NEXT: paddw %xmm1, %xmm0
+; SSE2-NEXT: movdqa %xmm0, %xmm1
+; SSE2-NEXT: psrld $16, %xmm1
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: movd %xmm1, %eax
+; SSE2-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v16i16_v16i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovsxbw %xmm0, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
+; SSE41-NEXT: pmovsxbw %xmm0, %xmm0
+; SSE41-NEXT: paddw %xmm1, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE41-NEXT: paddw %xmm1, %xmm0
+; SSE41-NEXT: movdqa %xmm0, %xmm1
+; SSE41-NEXT: psrld $16, %xmm1
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: movd %xmm1, %eax
+; SSE41-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v16i16_v16i8:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vpmovsxbw %xmm0, %xmm1
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpmovsxbw %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpaddw %xmm0, %xmm1, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v16i16_v16i8:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vpmovsxbw %xmm0, %xmm1
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
+; AVX1-FAST-NEXT: vpmovsxbw %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddw %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v16i16_v16i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovsxbw %xmm0, %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v16i16_v16i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovsxbw %xmm0, %ymm0
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %1 = sext <16 x i8> %a0 to <16 x i16>
+ %2 = call i16 @llvm.vector.reduce.add.v16i16(<16 x i16> %1)
+ ret i16 %2
+}
+
+define i16 @test_v32i16_v32i8(<32 x i8> %a0) {
+; SSE2-LABEL: test_v32i16_v32i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
+; SSE2-NEXT: psraw $8, %xmm2
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
+; SSE2-NEXT: psraw $8, %xmm3
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm1 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; SSE2-NEXT: psraw $8, %xmm1
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm0 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; SSE2-NEXT: psraw $8, %xmm0
+; SSE2-NEXT: paddw %xmm1, %xmm0
+; SSE2-NEXT: paddw %xmm2, %xmm0
+; SSE2-NEXT: paddw %xmm3, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE2-NEXT: paddw %xmm1, %xmm0
+; SSE2-NEXT: movdqa %xmm0, %xmm1
+; SSE2-NEXT: psrld $16, %xmm1
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: movd %xmm1, %eax
+; SSE2-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v32i16_v32i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm1[2,3,2,3]
+; SSE41-NEXT: pmovsxbw %xmm2, %xmm2
+; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm0[2,3,2,3]
+; SSE41-NEXT: pmovsxbw %xmm3, %xmm3
+; SSE41-NEXT: paddw %xmm2, %xmm3
+; SSE41-NEXT: pmovsxbw %xmm1, %xmm1
+; SSE41-NEXT: paddw %xmm3, %xmm1
+; SSE41-NEXT: pmovsxbw %xmm0, %xmm0
+; SSE41-NEXT: paddw %xmm1, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE41-NEXT: paddw %xmm1, %xmm0
+; SSE41-NEXT: movdqa %xmm0, %xmm1
+; SSE41-NEXT: psrld $16, %xmm1
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: movd %xmm1, %eax
+; SSE41-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v32i16_v32i8:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vextractf128 $1, %ymm0, %xmm1
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm2 = xmm1[2,3,2,3]
+; AVX1-SLOW-NEXT: vpmovsxbw %xmm2, %xmm2
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm3 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpmovsxbw %xmm3, %xmm3
+; AVX1-SLOW-NEXT: vpaddw %xmm2, %xmm3, %xmm2
+; AVX1-SLOW-NEXT: vpmovsxbw %xmm1, %xmm1
+; AVX1-SLOW-NEXT: vpaddw %xmm2, %xmm1, %xmm1
+; AVX1-SLOW-NEXT: vpmovsxbw %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-SLOW-NEXT: vzeroupper
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v32i16_v32i8:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vextractf128 $1, %ymm0, %xmm1
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm2 = xmm1[2,3,2,3]
+; AVX1-FAST-NEXT: vpmovsxbw %xmm2, %xmm2
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm3 = xmm0[2,3,2,3]
+; AVX1-FAST-NEXT: vpmovsxbw %xmm3, %xmm3
+; AVX1-FAST-NEXT: vpaddw %xmm2, %xmm3, %xmm2
+; AVX1-FAST-NEXT: vpmovsxbw %xmm1, %xmm1
+; AVX1-FAST-NEXT: vpaddw %xmm2, %xmm1, %xmm1
+; AVX1-FAST-NEXT: vpmovsxbw %xmm0, %xmm0
+; AVX1-FAST-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-FAST-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-FAST-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-FAST-NEXT: vzeroupper
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v32i16_v32i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpmovsxbw %xmm1, %ymm1
+; AVX2-NEXT: vpmovsxbw %xmm0, %ymm0
+; AVX2-NEXT: vpaddw %ymm1, %ymm0, %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v32i16_v32i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovsxbw %ymm0, %zmm0
+; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm1
+; AVX512-NEXT: vpaddw %zmm1, %zmm0, %zmm0
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %1 = sext <32 x i8> %a0 to <32 x i16>
+ %2 = call i16 @llvm.vector.reduce.add.v32i16(<32 x i16> %1)
+ ret i16 %2
+}
+
+define i16 @test_v64i16_v64i8(<64 x i8> %a0) {
+; SSE2-LABEL: test_v64i16_v64i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm4 = xmm4[8],xmm2[8],xmm4[9],xmm2[9],xmm4[10],xmm2[10],xmm4[11],xmm2[11],xmm4[12],xmm2[12],xmm4[13],xmm2[13],xmm4[14],xmm2[14],xmm4[15],xmm2[15]
+; SSE2-NEXT: psraw $8, %xmm4
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm5 = xmm5[8],xmm0[8],xmm5[9],xmm0[9],xmm5[10],xmm0[10],xmm5[11],xmm0[11],xmm5[12],xmm0[12],xmm5[13],xmm0[13],xmm5[14],xmm0[14],xmm5[15],xmm0[15]
+; SSE2-NEXT: psraw $8, %xmm5
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm6 = xmm6[8],xmm3[8],xmm6[9],xmm3[9],xmm6[10],xmm3[10],xmm6[11],xmm3[11],xmm6[12],xmm3[12],xmm6[13],xmm3[13],xmm6[14],xmm3[14],xmm6[15],xmm3[15]
+; SSE2-NEXT: psraw $8, %xmm6
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm7 = xmm7[8],xmm1[8],xmm7[9],xmm1[9],xmm7[10],xmm1[10],xmm7[11],xmm1[11],xmm7[12],xmm1[12],xmm7[13],xmm1[13],xmm7[14],xmm1[14],xmm7[15],xmm1[15]
+; SSE2-NEXT: psraw $8, %xmm7
+; SSE2-NEXT: paddw %xmm6, %xmm7
+; SSE2-NEXT: paddw %xmm4, %xmm7
+; SSE2-NEXT: paddw %xmm5, %xmm7
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE2-NEXT: psraw $8, %xmm2
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE2-NEXT: psraw $8, %xmm0
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm3 = xmm3[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE2-NEXT: psraw $8, %xmm3
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE2-NEXT: psraw $8, %xmm1
+; SSE2-NEXT: paddw %xmm3, %xmm1
+; SSE2-NEXT: paddw %xmm2, %xmm1
+; SSE2-NEXT: paddw %xmm7, %xmm1
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[2,3,2,3]
+; SSE2-NEXT: paddw %xmm1, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: movdqa %xmm1, %xmm0
+; SSE2-NEXT: psrld $16, %xmm0
+; SSE2-NEXT: paddw %xmm1, %xmm0
+; SSE2-NEXT: movd %xmm0, %eax
+; SSE2-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v64i16_v64i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovsxbw %xmm2, %xmm4
+; SSE41-NEXT: pmovsxbw %xmm0, %xmm5
+; SSE41-NEXT: pmovsxbw %xmm3, %xmm6
+; SSE41-NEXT: pmovsxbw %xmm1, %xmm7
+; SSE41-NEXT: paddw %xmm6, %xmm7
+; SSE41-NEXT: paddw %xmm4, %xmm7
+; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm2[2,3,2,3]
+; SSE41-NEXT: pmovsxbw %xmm2, %xmm2
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
+; SSE41-NEXT: pmovsxbw %xmm0, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm3[2,3,2,3]
+; SSE41-NEXT: pmovsxbw %xmm3, %xmm3
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm1[2,3,2,3]
+; SSE41-NEXT: pmovsxbw %xmm1, %xmm1
+; SSE41-NEXT: paddw %xmm3, %xmm1
+; SSE41-NEXT: paddw %xmm2, %xmm1
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: paddw %xmm7, %xmm1
+; SSE41-NEXT: paddw %xmm5, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[2,3,2,3]
+; SSE41-NEXT: paddw %xmm1, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: movdqa %xmm1, %xmm0
+; SSE41-NEXT: psrld $16, %xmm0
+; SSE41-NEXT: paddw %xmm1, %xmm0
+; SSE41-NEXT: movd %xmm0, %eax
+; SSE41-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v64i16_v64i8:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vpmovsxbw %xmm1, %xmm2
+; AVX1-SLOW-NEXT: vpmovsxbw %xmm0, %xmm3
+; AVX1-SLOW-NEXT: vextractf128 $1, %ymm1, %xmm4
+; AVX1-SLOW-NEXT: vpmovsxbw %xmm4, %xmm5
+; AVX1-SLOW-NEXT: vextractf128 $1, %ymm0, %xmm6
+; AVX1-SLOW-NEXT: vpmovsxbw %xmm6, %xmm7
+; AVX1-SLOW-NEXT: vpaddw %xmm5, %xmm7, %xmm5
+; AVX1-SLOW-NEXT: vpaddw %xmm5, %xmm2, %xmm2
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[2,3,2,3]
+; AVX1-SLOW-NEXT: vpmovsxbw %xmm1, %xmm1
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpmovsxbw %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm4 = xmm4[2,3,2,3]
+; AVX1-SLOW-NEXT: vpmovsxbw %xmm4, %xmm4
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm5 = xmm6[2,3,2,3]
+; AVX1-SLOW-NEXT: vpmovsxbw %xmm5, %xmm5
+; AVX1-SLOW-NEXT: vpaddw %xmm4, %xmm5, %xmm4
+; AVX1-SLOW-NEXT: vpaddw %xmm4, %xmm1, %xmm1
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpaddw %xmm0, %xmm2, %xmm0
+; AVX1-SLOW-NEXT: vpaddw %xmm0, %xmm3, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-SLOW-NEXT: vzeroupper
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v64i16_v64i8:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vpmovsxbw %xmm1, %xmm2
+; AVX1-FAST-NEXT: vpmovsxbw %xmm0, %xmm3
+; AVX1-FAST-NEXT: vextractf128 $1, %ymm1, %xmm4
+; AVX1-FAST-NEXT: vpmovsxbw %xmm4, %xmm5
+; AVX1-FAST-NEXT: vextractf128 $1, %ymm0, %xmm6
+; AVX1-FAST-NEXT: vpmovsxbw %xmm6, %xmm7
+; AVX1-FAST-NEXT: vpaddw %xmm5, %xmm7, %xmm5
+; AVX1-FAST-NEXT: vpaddw %xmm5, %xmm2, %xmm2
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[2,3,2,3]
+; AVX1-FAST-NEXT: vpmovsxbw %xmm1, %xmm1
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
+; AVX1-FAST-NEXT: vpmovsxbw %xmm0, %xmm0
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm4 = xmm4[2,3,2,3]
+; AVX1-FAST-NEXT: vpmovsxbw %xmm4, %xmm4
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm5 = xmm6[2,3,2,3]
+; AVX1-FAST-NEXT: vpmovsxbw %xmm5, %xmm5
+; AVX1-FAST-NEXT: vpaddw %xmm4, %xmm5, %xmm4
+; AVX1-FAST-NEXT: vpaddw %xmm4, %xmm1, %xmm1
+; AVX1-FAST-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vpaddw %xmm0, %xmm2, %xmm0
+; AVX1-FAST-NEXT: vpaddw %xmm0, %xmm3, %xmm0
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-FAST-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-FAST-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-FAST-NEXT: vzeroupper
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v64i16_v64i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovsxbw %xmm1, %ymm2
+; AVX2-NEXT: vpmovsxbw %xmm0, %ymm3
+; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm1
+; AVX2-NEXT: vpmovsxbw %xmm1, %ymm1
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm0
+; AVX2-NEXT: vpmovsxbw %xmm0, %ymm0
+; AVX2-NEXT: vpaddw %ymm1, %ymm0, %ymm0
+; AVX2-NEXT: vpaddw %ymm0, %ymm2, %ymm0
+; AVX2-NEXT: vpaddw %ymm0, %ymm3, %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v64i16_v64i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm1
+; AVX512-NEXT: vpmovsxbw %ymm1, %zmm1
+; AVX512-NEXT: vpmovsxbw %ymm0, %zmm0
+; AVX512-NEXT: vpaddw %zmm1, %zmm0, %zmm0
+; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm1
+; AVX512-NEXT: vpaddw %zmm1, %zmm0, %zmm0
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %1 = sext <64 x i8> %a0 to <64 x i16>
+ %2 = call i16 @llvm.vector.reduce.add.v64i16(<64 x i16> %1)
+ ret i16 %2
+
+}
+
+declare i64 @llvm.vector.reduce.add.v2i64(<2 x i64>)
+declare i64 @llvm.vector.reduce.add.v4i64(<4 x i64>)
+declare i64 @llvm.vector.reduce.add.v8i64(<8 x i64>)
+declare i64 @llvm.vector.reduce.add.v16i64(<16 x i64>)
+
+declare i32 @llvm.vector.reduce.add.v2i32(<2 x i32>)
+declare i32 @llvm.vector.reduce.add.v4i32(<4 x i32>)
+declare i32 @llvm.vector.reduce.add.v8i32(<8 x i32>)
+declare i32 @llvm.vector.reduce.add.v16i32(<16 x i32>)
+declare i32 @llvm.vector.reduce.add.v32i32(<32 x i32>)
+
+declare i16 @llvm.vector.reduce.add.v2i16(<2 x i16>)
+declare i16 @llvm.vector.reduce.add.v4i16(<4 x i16>)
+declare i16 @llvm.vector.reduce.add.v8i16(<8 x i16>)
+declare i16 @llvm.vector.reduce.add.v16i16(<16 x i16>)
+declare i16 @llvm.vector.reduce.add.v32i16(<32 x i16>)
+declare i16 @llvm.vector.reduce.add.v64i16(<64 x i16>)
+
+declare i8 @llvm.vector.reduce.add.v2i8(<2 x i8>)
+declare i8 @llvm.vector.reduce.add.v4i8(<4 x i8>)
+declare i8 @llvm.vector.reduce.add.v8i8(<8 x i8>)
+declare i8 @llvm.vector.reduce.add.v16i8(<16 x i8>)
+declare i8 @llvm.vector.reduce.add.v32i8(<32 x i8>)
+declare i8 @llvm.vector.reduce.add.v64i8(<64 x i8>)
+declare i8 @llvm.vector.reduce.add.v128i8(<128 x i8>)
diff --git a/llvm/test/CodeGen/X86/vector-reduce-add-zext.ll b/llvm/test/CodeGen/X86/vector-reduce-add-zext.ll
new file mode 100644
index 0000000000000..e9659ecab6689
--- /dev/null
+++ b/llvm/test/CodeGen/X86/vector-reduce-add-zext.ll
@@ -0,0 +1,1487 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefixes=SSE2
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse4.1 | FileCheck %s --check-prefixes=SSE41
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefixes=AVX,AVX1,AVX1-SLOW
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx,+fast-hops | FileCheck %s --check-prefixes=AVX,AVX1,AVX1-FAST
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefixes=AVX,AVX2
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512bw | FileCheck %s --check-prefix=AVX512
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512bw,+avx512vl | FileCheck %s --check-prefix=AVX512
+
+;
+; vXi64
+;
+
+define i64 @test_v2i64_v2i32(<2 x i32> %a0) {
+; SSE2-LABEL: test_v2i64_v2i32:
+; SSE2: # %bb.0:
+; SSE2-NEXT: xorps %xmm1, %xmm1
+; SSE2-NEXT: movss {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3]
+; SSE2-NEXT: psrlq $32, %xmm0
+; SSE2-NEXT: paddq %xmm1, %xmm0
+; SSE2-NEXT: movq %xmm0, %rax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v2i64_v2i32:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddq %xmm0, %xmm1
+; SSE41-NEXT: movq %xmm1, %rax
+; SSE41-NEXT: retq
+;
+; AVX-LABEL: test_v2i64_v2i32:
+; AVX: # %bb.0:
+; AVX-NEXT: vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
+; AVX-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX-NEXT: vmovq %xmm0, %rax
+; AVX-NEXT: retq
+;
+; AVX512-LABEL: test_v2i64_v2i32:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovq %xmm0, %rax
+; AVX512-NEXT: retq
+ %1 = zext <2 x i32> %a0 to <2 x i64>
+ %2 = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %1)
+ ret i64 %2
+}
+
+define i64 @test_v4i64_v4i16(<4 x i16> %a0) {
+; SSE2-LABEL: test_v4i64_v4i16:
+; SSE2: # %bb.0:
+; SSE2-NEXT: pxor %xmm1, %xmm1
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; SSE2-NEXT: movdqa %xmm0, %xmm2
+; SSE2-NEXT: punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm1[2],xmm2[3],xmm1[3]
+; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE2-NEXT: paddq %xmm2, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddq %xmm0, %xmm1
+; SSE2-NEXT: movq %xmm1, %rax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v4i64_v4i16:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovzxwq {{.*#+}} xmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
+; SSE41-NEXT: pmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
+; SSE41-NEXT: paddq %xmm1, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddq %xmm0, %xmm1
+; SSE41-NEXT: movq %xmm1, %rax
+; SSE41-NEXT: retq
+;
+; AVX1-LABEL: test_v4i64_v4i16:
+; AVX1: # %bb.0:
+; AVX1-NEXT: vpmovzxwq {{.*#+}} xmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
+; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
+; AVX1-NEXT: vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
+; AVX1-NEXT: vpaddq %xmm0, %xmm1, %xmm0
+; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vmovq %xmm0, %rax
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: test_v4i64_v4i16:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovzxwq {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovq %xmm0, %rax
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v4i64_v4i16:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovzxwq {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovq %xmm0, %rax
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %1 = zext <4 x i16> %a0 to <4 x i64>
+ %2 = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %1)
+ ret i64 %2
+}
+
+define i64 @test_v8i64_v8i8(<8 x i8> %a0) {
+; SSE2-LABEL: test_v8i64_v8i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: pxor %xmm1, %xmm1
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
+; SSE2-NEXT: movdqa %xmm0, %xmm2
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm2 = xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
+; SSE2-NEXT: movdqa %xmm2, %xmm3
+; SSE2-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; SSE2-NEXT: movdqa %xmm0, %xmm4
+; SSE2-NEXT: punpckldq {{.*#+}} xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1]
+; SSE2-NEXT: punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm1[2],xmm2[3],xmm1[3]
+; SSE2-NEXT: punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; SSE2-NEXT: paddq %xmm2, %xmm0
+; SSE2-NEXT: paddq %xmm3, %xmm0
+; SSE2-NEXT: paddq %xmm4, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddq %xmm0, %xmm1
+; SSE2-NEXT: movq %xmm1, %rax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v8i64_v8i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovzxbq {{.*#+}} xmm1 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
+; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm0[1,1,1,1]
+; SSE41-NEXT: pmovzxbq {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,zero,zero,zero,zero,xmm2[1],zero,zero,zero,zero,zero,zero,zero
+; SSE41-NEXT: movdqa %xmm0, %xmm3
+; SSE41-NEXT: psrlq $48, %xmm3
+; SSE41-NEXT: pmovzxbq {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,zero,zero,zero,zero,xmm3[1],zero,zero,zero,zero,zero,zero,zero
+; SSE41-NEXT: psrld $16, %xmm0
+; SSE41-NEXT: pmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
+; SSE41-NEXT: paddq %xmm3, %xmm0
+; SSE41-NEXT: paddq %xmm2, %xmm0
+; SSE41-NEXT: paddq %xmm1, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddq %xmm0, %xmm1
+; SSE41-NEXT: movq %xmm1, %rax
+; SSE41-NEXT: retq
+;
+; AVX1-LABEL: test_v8i64_v8i8:
+; AVX1: # %bb.0:
+; AVX1-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX1-NEXT: vpmovzxbq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,zero,zero,zero,zero,xmm1[1],zero,zero,zero,zero,zero,zero,zero
+; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm0[1,1,1,1]
+; AVX1-NEXT: vpsrld $16, %xmm2, %xmm3
+; AVX1-NEXT: vpmovzxbq {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,zero,zero,zero,zero,xmm3[1],zero,zero,zero,zero,zero,zero,zero
+; AVX1-NEXT: vpaddq %xmm3, %xmm1, %xmm1
+; AVX1-NEXT: vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
+; AVX1-NEXT: vpmovzxbq {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,zero,zero,zero,zero,xmm2[1],zero,zero,zero,zero,zero,zero,zero
+; AVX1-NEXT: vpaddq %xmm1, %xmm2, %xmm1
+; AVX1-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vmovq %xmm0, %rax
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: test_v8i64_v8i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovzxbq {{.*#+}} ymm1 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero,xmm0[2],zero,zero,zero,zero,zero,zero,zero,xmm0[3],zero,zero,zero,zero,zero,zero,zero
+; AVX2-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpmovzxbq {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero,xmm0[2],zero,zero,zero,zero,zero,zero,zero,xmm0[3],zero,zero,zero,zero,zero,zero,zero
+; AVX2-NEXT: vpaddq %ymm0, %ymm1, %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovq %xmm0, %rax
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v8i64_v8i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovzxbq {{.*#+}} zmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero,xmm0[2],zero,zero,zero,zero,zero,zero,zero,xmm0[3],zero,zero,zero,zero,zero,zero,zero,xmm0[4],zero,zero,zero,zero,zero,zero,zero,xmm0[5],zero,zero,zero,zero,zero,zero,zero,xmm0[6],zero,zero,zero,zero,zero,zero,zero,xmm0[7],zero,zero,zero,zero,zero,zero,zero
+; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm1
+; AVX512-NEXT: vpaddq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovq %xmm0, %rax
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %1 = zext <8 x i8> %a0 to <8 x i64>
+ %2 = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %1)
+ ret i64 %2
+}
+
+define i64 @test_v16i64_v16i8(<16 x i8> %a0) {
+; SSE2-LABEL: test_v16i64_v16i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: pxor %xmm3, %xmm3
+; SSE2-NEXT: movdqa %xmm0, %xmm1
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm3[8],xmm1[9],xmm3[9],xmm1[10],xmm3[10],xmm1[11],xmm3[11],xmm1[12],xmm3[12],xmm1[13],xmm3[13],xmm1[14],xmm3[14],xmm1[15],xmm3[15]
+; SSE2-NEXT: movdqa %xmm1, %xmm2
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3]
+; SSE2-NEXT: movdqa %xmm2, %xmm8
+; SSE2-NEXT: punpckhdq {{.*#+}} xmm8 = xmm8[2],xmm3[2],xmm8[3],xmm3[3]
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
+; SSE2-NEXT: movdqa %xmm0, %xmm4
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1],xmm4[2],xmm3[2],xmm4[3],xmm3[3]
+; SSE2-NEXT: movdqa %xmm4, %xmm6
+; SSE2-NEXT: punpckhdq {{.*#+}} xmm6 = xmm6[2],xmm3[2],xmm6[3],xmm3[3]
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm1 = xmm1[4],xmm3[4],xmm1[5],xmm3[5],xmm1[6],xmm3[6],xmm1[7],xmm3[7]
+; SSE2-NEXT: movdqa %xmm1, %xmm7
+; SSE2-NEXT: punpckhdq {{.*#+}} xmm7 = xmm7[2],xmm3[2],xmm7[3],xmm3[3]
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm0 = xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
+; SSE2-NEXT: movdqa %xmm0, %xmm5
+; SSE2-NEXT: punpckhdq {{.*#+}} xmm5 = xmm5[2],xmm3[2],xmm5[3],xmm3[3]
+; SSE2-NEXT: paddq %xmm7, %xmm5
+; SSE2-NEXT: paddq %xmm8, %xmm5
+; SSE2-NEXT: paddq %xmm6, %xmm5
+; SSE2-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
+; SSE2-NEXT: punpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
+; SSE2-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
+; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
+; SSE2-NEXT: paddq %xmm1, %xmm0
+; SSE2-NEXT: paddq %xmm2, %xmm0
+; SSE2-NEXT: paddq %xmm5, %xmm0
+; SSE2-NEXT: paddq %xmm4, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddq %xmm0, %xmm1
+; SSE2-NEXT: movq %xmm1, %rax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v16i64_v16i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: movdqa %xmm0, %xmm1
+; SSE41-NEXT: movdqa %xmm0, %xmm2
+; SSE41-NEXT: psrldq {{.*#+}} xmm2 = xmm2[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; SSE41-NEXT: pmovzxbq {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,zero,zero,zero,zero,xmm2[1],zero,zero,zero,zero,zero,zero,zero
+; SSE41-NEXT: movdqa %xmm0, %xmm3
+; SSE41-NEXT: psrlq $48, %xmm3
+; SSE41-NEXT: pmovzxbq {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,zero,zero,zero,zero,xmm3[1],zero,zero,zero,zero,zero,zero,zero
+; SSE41-NEXT: paddq %xmm2, %xmm3
+; SSE41-NEXT: pmovzxbq {{.*#+}} xmm2 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
+; SSE41-NEXT: pshufd {{.*#+}} xmm4 = xmm0[2,3,2,3]
+; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm0[3,3,3,3]
+; SSE41-NEXT: pshufd {{.*#+}} xmm6 = xmm0[1,1,1,1]
+; SSE41-NEXT: psrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; SSE41-NEXT: pmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
+; SSE41-NEXT: paddq %xmm0, %xmm3
+; SSE41-NEXT: psrld $16, %xmm1
+; SSE41-NEXT: pmovzxbq {{.*#+}} xmm0 = xmm1[0],zero,zero,zero,zero,zero,zero,zero,xmm1[1],zero,zero,zero,zero,zero,zero,zero
+; SSE41-NEXT: paddq %xmm0, %xmm3
+; SSE41-NEXT: pmovzxbq {{.*#+}} xmm0 = xmm4[0],zero,zero,zero,zero,zero,zero,zero,xmm4[1],zero,zero,zero,zero,zero,zero,zero
+; SSE41-NEXT: pmovzxbq {{.*#+}} xmm1 = xmm5[0],zero,zero,zero,zero,zero,zero,zero,xmm5[1],zero,zero,zero,zero,zero,zero,zero
+; SSE41-NEXT: pmovzxbq {{.*#+}} xmm4 = xmm6[0],zero,zero,zero,zero,zero,zero,zero,xmm6[1],zero,zero,zero,zero,zero,zero,zero
+; SSE41-NEXT: paddq %xmm1, %xmm4
+; SSE41-NEXT: paddq %xmm0, %xmm4
+; SSE41-NEXT: paddq %xmm3, %xmm4
+; SSE41-NEXT: paddq %xmm2, %xmm4
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm4[2,3,2,3]
+; SSE41-NEXT: paddq %xmm4, %xmm0
+; SSE41-NEXT: movq %xmm0, %rax
+; SSE41-NEXT: retq
+;
+; AVX1-LABEL: test_v16i64_v16i8:
+; AVX1: # %bb.0:
+; AVX1-NEXT: vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT: vpunpckhbw {{.*#+}} xmm1 = xmm0[8],xmm1[8],xmm0[9],xmm1[9],xmm0[10],xmm1[10],xmm0[11],xmm1[11],xmm0[12],xmm1[12],xmm0[13],xmm1[13],xmm0[14],xmm1[14],xmm0[15],xmm1[15]
+; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm1[1,1,1,1]
+; AVX1-NEXT: vpmovzxwq {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero
+; AVX1-NEXT: vpshufd {{.*#+}} xmm3 = xmm1[3,3,3,3]
+; AVX1-NEXT: vpmovzxwq {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero
+; AVX1-NEXT: vpshufd {{.*#+}} xmm4 = xmm1[2,3,2,3]
+; AVX1-NEXT: vpmovzxwq {{.*#+}} xmm4 = xmm4[0],zero,zero,zero,xmm4[1],zero,zero,zero
+; AVX1-NEXT: vpmovzxwq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero
+; AVX1-NEXT: vpsrld $16, %xmm0, %xmm5
+; AVX1-NEXT: vpmovzxbq {{.*#+}} xmm5 = xmm5[0],zero,zero,zero,zero,zero,zero,zero,xmm5[1],zero,zero,zero,zero,zero,zero,zero
+; AVX1-NEXT: vpmovzxbw {{.*#+}} xmm6 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX1-NEXT: vpshufd {{.*#+}} xmm7 = xmm6[3,3,3,3]
+; AVX1-NEXT: vpmovzxwq {{.*#+}} xmm7 = xmm7[0],zero,zero,zero,xmm7[1],zero,zero,zero
+; AVX1-NEXT: vpaddq %xmm3, %xmm7, %xmm3
+; AVX1-NEXT: vpaddq %xmm3, %xmm2, %xmm2
+; AVX1-NEXT: vpaddq %xmm2, %xmm5, %xmm2
+; AVX1-NEXT: vpshufd {{.*#+}} xmm3 = xmm6[2,3,2,3]
+; AVX1-NEXT: vpmovzxwq {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero
+; AVX1-NEXT: vpaddq %xmm4, %xmm3, %xmm3
+; AVX1-NEXT: vpaddq %xmm3, %xmm1, %xmm1
+; AVX1-NEXT: vpaddq %xmm2, %xmm1, %xmm1
+; AVX1-NEXT: vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
+; AVX1-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vmovq %xmm0, %rax
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: test_v16i64_v16i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovzxbw {{.*#+}} ymm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
+; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm2
+; AVX2-NEXT: vpshufd {{.*#+}} xmm3 = xmm2[2,3,2,3]
+; AVX2-NEXT: vpmovzxwq {{.*#+}} ymm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[2,3,2,3]
+; AVX2-NEXT: vpmovzxwq {{.*#+}} ymm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero
+; AVX2-NEXT: vpaddq %ymm3, %ymm1, %ymm1
+; AVX2-NEXT: vpmovzxbq {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero,xmm0[2],zero,zero,zero,zero,zero,zero,zero,xmm0[3],zero,zero,zero,zero,zero,zero,zero
+; AVX2-NEXT: vpmovzxwq {{.*#+}} ymm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
+; AVX2-NEXT: vpaddq %ymm1, %ymm2, %ymm1
+; AVX2-NEXT: vpaddq %ymm1, %ymm0, %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovq %xmm0, %rax
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v16i64_v16i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpmovzxwq {{.*#+}} zmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero
+; AVX512-NEXT: vpmovzxwq {{.*#+}} zmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero
+; AVX512-NEXT: vpaddq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm1
+; AVX512-NEXT: vpaddq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddq %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovq %xmm0, %rax
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %1 = zext <16 x i8> %a0 to <16 x i64>
+ %2 = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %1)
+ ret i64 %2
+}
+
+;
+; vXi32
+;
+
+define i32 @test_v2i32_v2i16(<2 x i16> %a0) {
+; SSE2-LABEL: test_v2i32_v2i16:
+; SSE2: # %bb.0:
+; SSE2-NEXT: pxor %xmm1, %xmm1
+; SSE2-NEXT: movdqa %xmm0, %xmm2
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
+; SSE2-NEXT: psrld $16, %xmm0
+; SSE2-NEXT: paddd %xmm2, %xmm0
+; SSE2-NEXT: movd %xmm0, %eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v2i32_v2i16:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovzxwd {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
+; SSE41-NEXT: psrld $16, %xmm0
+; SSE41-NEXT: paddd %xmm1, %xmm0
+; SSE41-NEXT: movd %xmm0, %eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v2i32_v2i16:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vpmovzxwd {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
+; AVX1-SLOW-NEXT: vpsrld $16, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpaddd %xmm0, %xmm1, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v2i32_v2i16:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
+; AVX1-FAST-NEXT: vphaddd %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v2i32_v2i16:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovzxwd {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
+; AVX2-NEXT: vpsrld $16, %xmm0, %xmm0
+; AVX2-NEXT: vpaddd %xmm0, %xmm1, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v2i32_v2i16:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovzxwd {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
+; AVX512-NEXT: vpsrld $16, %xmm0, %xmm0
+; AVX512-NEXT: vpaddd %xmm0, %xmm1, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: retq
+ %1 = zext <2 x i16> %a0 to <2 x i32>
+ %2 = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %1)
+ ret i32 %2
+}
+
+define i32 @test_v4i32(<4 x i8> %a0) {
+; SSE2-LABEL: test_v4i32:
+; SSE2: # %bb.0:
+; SSE2-NEXT: pxor %xmm1, %xmm1
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddd %xmm0, %xmm1
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE2-NEXT: paddd %xmm1, %xmm0
+; SSE2-NEXT: movd %xmm0, %eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v4i32:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddd %xmm0, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE41-NEXT: paddd %xmm1, %xmm0
+; SSE41-NEXT: movd %xmm0, %eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v4i32:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v4i32:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; AVX1-FAST-NEXT: vphaddd %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddd %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v4i32:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v4i32:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: retq
+ %1 = zext <4 x i8> %a0 to <4 x i32>
+ %2 = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %1)
+ ret i32 %2
+}
+
+define i32 @test_v8i32_v8i8(<8 x i8> %a0) {
+; SSE2-LABEL: test_v8i32_v8i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: pxor %xmm1, %xmm1
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
+; SSE2-NEXT: movdqa %xmm0, %xmm2
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm2 = xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; SSE2-NEXT: paddd %xmm2, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddd %xmm0, %xmm1
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE2-NEXT: paddd %xmm1, %xmm0
+; SSE2-NEXT: movd %xmm0, %eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v8i32_v8i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovzxbd {{.*#+}} xmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
+; SSE41-NEXT: pmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; SSE41-NEXT: paddd %xmm1, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddd %xmm0, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE41-NEXT: paddd %xmm1, %xmm0
+; SSE41-NEXT: movd %xmm0, %eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v8i32_v8i8:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vpmovzxbd {{.*#+}} xmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; AVX1-SLOW-NEXT: vpaddd %xmm0, %xmm1, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v8i32_v8i8:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vpmovzxbd {{.*#+}} xmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
+; AVX1-FAST-NEXT: vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; AVX1-FAST-NEXT: vphaddd %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddd %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddd %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v8i32_v8i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovzxbd {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v8i32_v8i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovzxbd {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %1 = zext <8 x i8> %a0 to <8 x i32>
+ %2 = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %1)
+ ret i32 %2
+}
+
+define i32 @test_v16i32_v16i8(<16 x i8> %a0) {
+; SSE2-LABEL: test_v16i32_v16i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: pxor %xmm1, %xmm1
+; SSE2-NEXT: movdqa %xmm0, %xmm2
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm1[8],xmm2[9],xmm1[9],xmm2[10],xmm1[10],xmm2[11],xmm1[11],xmm2[12],xmm1[12],xmm2[13],xmm1[13],xmm2[14],xmm1[14],xmm2[15],xmm1[15]
+; SSE2-NEXT: movdqa %xmm2, %xmm3
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3]
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
+; SSE2-NEXT: movdqa %xmm0, %xmm4
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1],xmm4[2],xmm1[2],xmm4[3],xmm1[3]
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm2 = xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
+; SSE2-NEXT: paddd %xmm2, %xmm0
+; SSE2-NEXT: paddd %xmm3, %xmm0
+; SSE2-NEXT: paddd %xmm4, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddd %xmm0, %xmm1
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE2-NEXT: paddd %xmm1, %xmm0
+; SSE2-NEXT: movd %xmm0, %eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v16i32_v16i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovzxbd {{.*#+}} xmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm0[2,3,2,3]
+; SSE41-NEXT: pmovzxbd {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
+; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm0[3,3,3,3]
+; SSE41-NEXT: pmovzxbd {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
+; SSE41-NEXT: pmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; SSE41-NEXT: paddd %xmm3, %xmm0
+; SSE41-NEXT: paddd %xmm2, %xmm0
+; SSE41-NEXT: paddd %xmm1, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddd %xmm0, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE41-NEXT: paddd %xmm1, %xmm0
+; SSE41-NEXT: movd %xmm0, %eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v16i32_v16i8:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vpmovzxbd {{.*#+}} xmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm2 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpmovzxbd {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm3 = xmm0[3,3,3,3]
+; AVX1-SLOW-NEXT: vpmovzxbd {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; AVX1-SLOW-NEXT: vpaddd %xmm3, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpaddd %xmm0, %xmm2, %xmm0
+; AVX1-SLOW-NEXT: vpaddd %xmm0, %xmm1, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v16i32_v16i8:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vpmovzxbd {{.*#+}} xmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm2 = xmm0[2,3,2,3]
+; AVX1-FAST-NEXT: vpmovzxbd {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm3 = xmm0[3,3,3,3]
+; AVX1-FAST-NEXT: vpmovzxbd {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
+; AVX1-FAST-NEXT: vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; AVX1-FAST-NEXT: vpaddd %xmm3, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vpaddd %xmm0, %xmm2, %xmm0
+; AVX1-FAST-NEXT: vpaddd %xmm0, %xmm1, %xmm0
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-FAST-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddd %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v16i32_v16i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovzxbd {{.*#+}} ymm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero
+; AVX2-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpmovzxbd {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero
+; AVX2-NEXT: vpaddd %ymm0, %ymm1, %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v16i32_v16i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovzxbd {{.*#+}} zmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
+; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm1
+; AVX512-NEXT: vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %1 = zext <16 x i8> %a0 to <16 x i32>
+ %2 = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %1)
+ ret i32 %2
+}
+
+define i32 @test_v32i32_v32i8(<32 x i8> %a0) {
+; SSE2-LABEL: test_v32i32_v32i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: pxor %xmm4, %xmm4
+; SSE2-NEXT: movdqa %xmm1, %xmm2
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1],xmm2[2],xmm4[2],xmm2[3],xmm4[3],xmm2[4],xmm4[4],xmm2[5],xmm4[5],xmm2[6],xmm4[6],xmm2[7],xmm4[7]
+; SSE2-NEXT: movdqa %xmm2, %xmm8
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm8 = xmm8[4],xmm4[4],xmm8[5],xmm4[5],xmm8[6],xmm4[6],xmm8[7],xmm4[7]
+; SSE2-NEXT: movdqa %xmm0, %xmm3
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3],xmm3[4],xmm4[4],xmm3[5],xmm4[5],xmm3[6],xmm4[6],xmm3[7],xmm4[7]
+; SSE2-NEXT: movdqa %xmm3, %xmm6
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm6 = xmm6[4],xmm4[4],xmm6[5],xmm4[5],xmm6[6],xmm4[6],xmm6[7],xmm4[7]
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm4[8],xmm1[9],xmm4[9],xmm1[10],xmm4[10],xmm1[11],xmm4[11],xmm1[12],xmm4[12],xmm1[13],xmm4[13],xmm1[14],xmm4[14],xmm1[15],xmm4[15]
+; SSE2-NEXT: movdqa %xmm1, %xmm7
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm7 = xmm7[4],xmm4[4],xmm7[5],xmm4[5],xmm7[6],xmm4[6],xmm7[7],xmm4[7]
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm4[8],xmm0[9],xmm4[9],xmm0[10],xmm4[10],xmm0[11],xmm4[11],xmm0[12],xmm4[12],xmm0[13],xmm4[13],xmm0[14],xmm4[14],xmm0[15],xmm4[15]
+; SSE2-NEXT: movdqa %xmm0, %xmm5
+; SSE2-NEXT: punpckhwd {{.*#+}} xmm5 = xmm5[4],xmm4[4],xmm5[5],xmm4[5],xmm5[6],xmm4[6],xmm5[7],xmm4[7]
+; SSE2-NEXT: paddd %xmm7, %xmm5
+; SSE2-NEXT: paddd %xmm8, %xmm5
+; SSE2-NEXT: paddd %xmm6, %xmm5
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1],xmm2[2],xmm4[2],xmm2[3],xmm4[3]
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3]
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1],xmm1[2],xmm4[2],xmm1[3],xmm4[3]
+; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3]
+; SSE2-NEXT: paddd %xmm1, %xmm0
+; SSE2-NEXT: paddd %xmm2, %xmm0
+; SSE2-NEXT: paddd %xmm5, %xmm0
+; SSE2-NEXT: paddd %xmm3, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddd %xmm0, %xmm1
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE2-NEXT: paddd %xmm1, %xmm0
+; SSE2-NEXT: movd %xmm0, %eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v32i32_v32i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm1[1,1,1,1]
+; SSE41-NEXT: pmovzxbd {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
+; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm0[1,1,1,1]
+; SSE41-NEXT: pmovzxbd {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
+; SSE41-NEXT: pshufd {{.*#+}} xmm4 = xmm1[3,3,3,3]
+; SSE41-NEXT: pmovzxbd {{.*#+}} xmm4 = xmm4[0],zero,zero,zero,xmm4[1],zero,zero,zero,xmm4[2],zero,zero,zero,xmm4[3],zero,zero,zero
+; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm0[3,3,3,3]
+; SSE41-NEXT: pmovzxbd {{.*#+}} xmm5 = xmm5[0],zero,zero,zero,xmm5[1],zero,zero,zero,xmm5[2],zero,zero,zero,xmm5[3],zero,zero,zero
+; SSE41-NEXT: paddd %xmm4, %xmm5
+; SSE41-NEXT: paddd %xmm2, %xmm5
+; SSE41-NEXT: paddd %xmm3, %xmm5
+; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm1[2,3,2,3]
+; SSE41-NEXT: pmovzxbd {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
+; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm0[2,3,2,3]
+; SSE41-NEXT: pmovzxbd {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
+; SSE41-NEXT: paddd %xmm2, %xmm3
+; SSE41-NEXT: pmovzxbd {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero
+; SSE41-NEXT: paddd %xmm3, %xmm1
+; SSE41-NEXT: paddd %xmm5, %xmm1
+; SSE41-NEXT: pmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; SSE41-NEXT: paddd %xmm1, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddd %xmm0, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE41-NEXT: paddd %xmm1, %xmm0
+; SSE41-NEXT: movd %xmm0, %eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v32i32_v32i8:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vextractf128 $1, %ymm0, %xmm1
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm2 = xmm1[1,1,1,1]
+; AVX1-SLOW-NEXT: vpmovzxbd {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm3 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpmovzxbd {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm4 = xmm1[3,3,3,3]
+; AVX1-SLOW-NEXT: vpmovzxbd {{.*#+}} xmm4 = xmm4[0],zero,zero,zero,xmm4[1],zero,zero,zero,xmm4[2],zero,zero,zero,xmm4[3],zero,zero,zero
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm5 = xmm0[3,3,3,3]
+; AVX1-SLOW-NEXT: vpmovzxbd {{.*#+}} xmm5 = xmm5[0],zero,zero,zero,xmm5[1],zero,zero,zero,xmm5[2],zero,zero,zero,xmm5[3],zero,zero,zero
+; AVX1-SLOW-NEXT: vpaddd %xmm4, %xmm5, %xmm4
+; AVX1-SLOW-NEXT: vpaddd %xmm4, %xmm2, %xmm2
+; AVX1-SLOW-NEXT: vpaddd %xmm2, %xmm3, %xmm2
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm3 = xmm1[2,3,2,3]
+; AVX1-SLOW-NEXT: vpmovzxbd {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm4 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpmovzxbd {{.*#+}} xmm4 = xmm4[0],zero,zero,zero,xmm4[1],zero,zero,zero,xmm4[2],zero,zero,zero,xmm4[3],zero,zero,zero
+; AVX1-SLOW-NEXT: vpaddd %xmm3, %xmm4, %xmm3
+; AVX1-SLOW-NEXT: vpmovzxbd {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero
+; AVX1-SLOW-NEXT: vpaddd %xmm3, %xmm1, %xmm1
+; AVX1-SLOW-NEXT: vpaddd %xmm2, %xmm1, %xmm1
+; AVX1-SLOW-NEXT: vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; AVX1-SLOW-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: vzeroupper
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v32i32_v32i8:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vextractf128 $1, %ymm0, %xmm1
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm2 = xmm1[1,1,1,1]
+; AVX1-FAST-NEXT: vpmovzxbd {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm3 = xmm0[1,1,1,1]
+; AVX1-FAST-NEXT: vpmovzxbd {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm4 = xmm1[3,3,3,3]
+; AVX1-FAST-NEXT: vpmovzxbd {{.*#+}} xmm4 = xmm4[0],zero,zero,zero,xmm4[1],zero,zero,zero,xmm4[2],zero,zero,zero,xmm4[3],zero,zero,zero
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm5 = xmm0[3,3,3,3]
+; AVX1-FAST-NEXT: vpmovzxbd {{.*#+}} xmm5 = xmm5[0],zero,zero,zero,xmm5[1],zero,zero,zero,xmm5[2],zero,zero,zero,xmm5[3],zero,zero,zero
+; AVX1-FAST-NEXT: vpaddd %xmm4, %xmm5, %xmm4
+; AVX1-FAST-NEXT: vpaddd %xmm4, %xmm2, %xmm2
+; AVX1-FAST-NEXT: vpaddd %xmm2, %xmm3, %xmm2
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm3 = xmm1[2,3,2,3]
+; AVX1-FAST-NEXT: vpmovzxbd {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm4 = xmm0[2,3,2,3]
+; AVX1-FAST-NEXT: vpmovzxbd {{.*#+}} xmm4 = xmm4[0],zero,zero,zero,xmm4[1],zero,zero,zero,xmm4[2],zero,zero,zero,xmm4[3],zero,zero,zero
+; AVX1-FAST-NEXT: vpaddd %xmm3, %xmm4, %xmm3
+; AVX1-FAST-NEXT: vpmovzxbd {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero
+; AVX1-FAST-NEXT: vpaddd %xmm3, %xmm1, %xmm1
+; AVX1-FAST-NEXT: vpaddd %xmm2, %xmm1, %xmm1
+; AVX1-FAST-NEXT: vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; AVX1-FAST-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-FAST-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddd %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: vzeroupper
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v32i32_v32i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpshufd {{.*#+}} xmm2 = xmm1[2,3,2,3]
+; AVX2-NEXT: vpmovzxbd {{.*#+}} ymm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero,xmm2[4],zero,zero,zero,xmm2[5],zero,zero,zero,xmm2[6],zero,zero,zero,xmm2[7],zero,zero,zero
+; AVX2-NEXT: vpshufd {{.*#+}} xmm3 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpmovzxbd {{.*#+}} ymm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero,xmm3[4],zero,zero,zero,xmm3[5],zero,zero,zero,xmm3[6],zero,zero,zero,xmm3[7],zero,zero,zero
+; AVX2-NEXT: vpaddd %ymm2, %ymm3, %ymm2
+; AVX2-NEXT: vpmovzxbd {{.*#+}} ymm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero
+; AVX2-NEXT: vpaddd %ymm2, %ymm1, %ymm1
+; AVX2-NEXT: vpmovzxbd {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero
+; AVX2-NEXT: vpaddd %ymm1, %ymm0, %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v32i32_v32i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpmovzxbd {{.*#+}} zmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero
+; AVX512-NEXT: vpmovzxbd {{.*#+}} zmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
+; AVX512-NEXT: vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm1
+; AVX512-NEXT: vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpaddd %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %1 = zext <32 x i8> %a0 to <32 x i32>
+ %2 = call i32 @llvm.vector.reduce.add.v32i32(<32 x i32> %1)
+ ret i32 %2
+}
+
+;
+; vXi16
+;
+
+define i16 @test_v2i16_v2i8(<2 x i8> %a0) {
+; SSE2-LABEL: test_v2i16_v2i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: pxor %xmm1, %xmm1
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
+; SSE2-NEXT: movdqa %xmm0, %xmm1
+; SSE2-NEXT: psrld $16, %xmm1
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: movd %xmm1, %eax
+; SSE2-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v2i16_v2i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; SSE41-NEXT: movdqa %xmm0, %xmm1
+; SSE41-NEXT: psrld $16, %xmm1
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: movd %xmm1, %eax
+; SSE41-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v2i16_v2i8:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX1-SLOW-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v2i16_v2i8:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v2i16_v2i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX2-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v2i16_v2i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX512-NEXT: retq
+ %1 = zext <2 x i8> %a0 to <2 x i16>
+ %2 = call i16 @llvm.vector.reduce.add.v2i16(<2 x i16> %1)
+ ret i16 %2
+}
+
+define i16 @test_v4i16_v4i8(<4 x i8> %a0) {
+; SSE2-LABEL: test_v4i16_v4i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: pxor %xmm1, %xmm1
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: movdqa %xmm1, %xmm0
+; SSE2-NEXT: psrld $16, %xmm0
+; SSE2-NEXT: paddw %xmm1, %xmm0
+; SSE2-NEXT: movd %xmm0, %eax
+; SSE2-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v4i16_v4i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: movdqa %xmm1, %xmm0
+; SSE41-NEXT: psrld $16, %xmm0
+; SSE41-NEXT: paddw %xmm1, %xmm0
+; SSE41-NEXT: movd %xmm0, %eax
+; SSE41-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v4i16_v4i8:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v4i16_v4i8:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-FAST-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v4i16_v4i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v4i16_v4i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX512-NEXT: retq
+ %1 = zext <4 x i8> %a0 to <4 x i16>
+ %2 = call i16 @llvm.vector.reduce.add.v4i16(<4 x i16> %1)
+ ret i16 %2
+
+}
+
+define i16 @test_v8i16_v8i8(<8 x i8> %a0) {
+; SSE2-LABEL: test_v8i16_v8i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: pxor %xmm1, %xmm1
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE2-NEXT: paddw %xmm1, %xmm0
+; SSE2-NEXT: movdqa %xmm0, %xmm1
+; SSE2-NEXT: psrld $16, %xmm1
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: movd %xmm1, %eax
+; SSE2-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v8i16_v8i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE41-NEXT: paddw %xmm1, %xmm0
+; SSE41-NEXT: movdqa %xmm0, %xmm1
+; SSE41-NEXT: psrld $16, %xmm1
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: movd %xmm1, %eax
+; SSE41-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v8i16_v8i8:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v8i16_v8i8:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v8i16_v8i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v8i16_v8i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX512-NEXT: retq
+ %1 = zext <8 x i8> %a0 to <8 x i16>
+ %2 = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %1)
+ ret i16 %2
+}
+
+define i16 @test_v16i16_v16i8(<16 x i8> %a0) {
+; SSE2-LABEL: test_v16i16_v16i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: pxor %xmm1, %xmm1
+; SSE2-NEXT: movdqa %xmm0, %xmm2
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm1[8],xmm2[9],xmm1[9],xmm2[10],xmm1[10],xmm2[11],xmm1[11],xmm2[12],xmm1[12],xmm2[13],xmm1[13],xmm2[14],xmm1[14],xmm2[15],xmm1[15]
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
+; SSE2-NEXT: paddw %xmm2, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE2-NEXT: paddw %xmm1, %xmm0
+; SSE2-NEXT: movdqa %xmm0, %xmm1
+; SSE2-NEXT: psrld $16, %xmm1
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: movd %xmm1, %eax
+; SSE2-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v16i16_v16i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pxor %xmm1, %xmm1
+; SSE41-NEXT: pmovzxbw {{.*#+}} xmm2 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; SSE41-NEXT: punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm1[8],xmm0[9],xmm1[9],xmm0[10],xmm1[10],xmm0[11],xmm1[11],xmm0[12],xmm1[12],xmm0[13],xmm1[13],xmm0[14],xmm1[14],xmm0[15],xmm1[15]
+; SSE41-NEXT: paddw %xmm2, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE41-NEXT: paddw %xmm1, %xmm0
+; SSE41-NEXT: movdqa %xmm0, %xmm1
+; SSE41-NEXT: psrld $16, %xmm1
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: movd %xmm1, %eax
+; SSE41-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v16i16_v16i8:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vpxor %xmm1, %xmm1, %xmm1
+; AVX1-SLOW-NEXT: vpunpckhbw {{.*#+}} xmm1 = xmm0[8],xmm1[8],xmm0[9],xmm1[9],xmm0[10],xmm1[10],xmm0[11],xmm1[11],xmm0[12],xmm1[12],xmm0[13],xmm1[13],xmm0[14],xmm1[14],xmm0[15],xmm1[15]
+; AVX1-SLOW-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v16i16_v16i8:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vpxor %xmm1, %xmm1, %xmm1
+; AVX1-FAST-NEXT: vpunpckhbw {{.*#+}} xmm1 = xmm0[8],xmm1[8],xmm0[9],xmm1[9],xmm0[10],xmm1[10],xmm0[11],xmm1[11],xmm0[12],xmm1[12],xmm0[13],xmm1[13],xmm0[14],xmm1[14],xmm0[15],xmm1[15]
+; AVX1-FAST-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm1, %xmm0
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v16i16_v16i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v16i16_v16i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %1 = zext <16 x i8> %a0 to <16 x i16>
+ %2 = call i16 @llvm.vector.reduce.add.v16i16(<16 x i16> %1)
+ ret i16 %2
+}
+
+define i16 @test_v32i16_v32i8(<32 x i8> %a0) {
+; SSE2-LABEL: test_v32i16_v32i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: pxor %xmm2, %xmm2
+; SSE2-NEXT: movdqa %xmm1, %xmm3
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3],xmm3[4],xmm2[4],xmm3[5],xmm2[5],xmm3[6],xmm2[6],xmm3[7],xmm2[7]
+; SSE2-NEXT: movdqa %xmm0, %xmm4
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm2[0],xmm4[1],xmm2[1],xmm4[2],xmm2[2],xmm4[3],xmm2[3],xmm4[4],xmm2[4],xmm4[5],xmm2[5],xmm4[6],xmm2[6],xmm4[7],xmm2[7]
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm2[8],xmm1[9],xmm2[9],xmm1[10],xmm2[10],xmm1[11],xmm2[11],xmm1[12],xmm2[12],xmm1[13],xmm2[13],xmm1[14],xmm2[14],xmm1[15],xmm2[15]
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm2[8],xmm0[9],xmm2[9],xmm0[10],xmm2[10],xmm0[11],xmm2[11],xmm0[12],xmm2[12],xmm0[13],xmm2[13],xmm0[14],xmm2[14],xmm0[15],xmm2[15]
+; SSE2-NEXT: paddw %xmm1, %xmm0
+; SSE2-NEXT: paddw %xmm3, %xmm0
+; SSE2-NEXT: paddw %xmm4, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE2-NEXT: paddw %xmm1, %xmm0
+; SSE2-NEXT: movdqa %xmm0, %xmm1
+; SSE2-NEXT: psrld $16, %xmm1
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: movd %xmm1, %eax
+; SSE2-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v32i16_v32i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pxor %xmm2, %xmm2
+; SSE41-NEXT: pmovzxbw {{.*#+}} xmm3 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
+; SSE41-NEXT: punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm2[8],xmm1[9],xmm2[9],xmm1[10],xmm2[10],xmm1[11],xmm2[11],xmm1[12],xmm2[12],xmm1[13],xmm2[13],xmm1[14],xmm2[14],xmm1[15],xmm2[15]
+; SSE41-NEXT: pmovzxbw {{.*#+}} xmm4 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; SSE41-NEXT: punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm2[8],xmm0[9],xmm2[9],xmm0[10],xmm2[10],xmm0[11],xmm2[11],xmm0[12],xmm2[12],xmm0[13],xmm2[13],xmm0[14],xmm2[14],xmm0[15],xmm2[15]
+; SSE41-NEXT: paddw %xmm1, %xmm0
+; SSE41-NEXT: paddw %xmm3, %xmm0
+; SSE41-NEXT: paddw %xmm4, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
+; SSE41-NEXT: paddw %xmm1, %xmm0
+; SSE41-NEXT: movdqa %xmm0, %xmm1
+; SSE41-NEXT: psrld $16, %xmm1
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: movd %xmm1, %eax
+; SSE41-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v32i16_v32i8:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vextractf128 $1, %ymm0, %xmm1
+; AVX1-SLOW-NEXT: vpxor %xmm2, %xmm2, %xmm2
+; AVX1-SLOW-NEXT: vpunpckhbw {{.*#+}} xmm3 = xmm1[8],xmm2[8],xmm1[9],xmm2[9],xmm1[10],xmm2[10],xmm1[11],xmm2[11],xmm1[12],xmm2[12],xmm1[13],xmm2[13],xmm1[14],xmm2[14],xmm1[15],xmm2[15]
+; AVX1-SLOW-NEXT: vpunpckhbw {{.*#+}} xmm2 = xmm0[8],xmm2[8],xmm0[9],xmm2[9],xmm0[10],xmm2[10],xmm0[11],xmm2[11],xmm0[12],xmm2[12],xmm0[13],xmm2[13],xmm0[14],xmm2[14],xmm0[15],xmm2[15]
+; AVX1-SLOW-NEXT: vpaddw %xmm3, %xmm2, %xmm2
+; AVX1-SLOW-NEXT: vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
+; AVX1-SLOW-NEXT: vpaddw %xmm2, %xmm1, %xmm1
+; AVX1-SLOW-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-SLOW-NEXT: vzeroupper
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v32i16_v32i8:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vextractf128 $1, %ymm0, %xmm1
+; AVX1-FAST-NEXT: vpxor %xmm2, %xmm2, %xmm2
+; AVX1-FAST-NEXT: vpunpckhbw {{.*#+}} xmm3 = xmm1[8],xmm2[8],xmm1[9],xmm2[9],xmm1[10],xmm2[10],xmm1[11],xmm2[11],xmm1[12],xmm2[12],xmm1[13],xmm2[13],xmm1[14],xmm2[14],xmm1[15],xmm2[15]
+; AVX1-FAST-NEXT: vpunpckhbw {{.*#+}} xmm2 = xmm0[8],xmm2[8],xmm0[9],xmm2[9],xmm0[10],xmm2[10],xmm0[11],xmm2[11],xmm0[12],xmm2[12],xmm0[13],xmm2[13],xmm0[14],xmm2[14],xmm0[15],xmm2[15]
+; AVX1-FAST-NEXT: vpaddw %xmm3, %xmm2, %xmm2
+; AVX1-FAST-NEXT: vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
+; AVX1-FAST-NEXT: vpaddw %xmm2, %xmm1, %xmm1
+; AVX1-FAST-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX1-FAST-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-FAST-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-FAST-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-FAST-NEXT: vzeroupper
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v32i16_v32i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpmovzxbw {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero,xmm1[8],zero,xmm1[9],zero,xmm1[10],zero,xmm1[11],zero,xmm1[12],zero,xmm1[13],zero,xmm1[14],zero,xmm1[15],zero
+; AVX2-NEXT: vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
+; AVX2-NEXT: vpaddw %ymm1, %ymm0, %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v32i16_v32i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vpmovzxbw {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero,ymm0[16],zero,ymm0[17],zero,ymm0[18],zero,ymm0[19],zero,ymm0[20],zero,ymm0[21],zero,ymm0[22],zero,ymm0[23],zero,ymm0[24],zero,ymm0[25],zero,ymm0[26],zero,ymm0[27],zero,ymm0[28],zero,ymm0[29],zero,ymm0[30],zero,ymm0[31],zero
+; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm1
+; AVX512-NEXT: vpaddw %zmm1, %zmm0, %zmm0
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %1 = zext <32 x i8> %a0 to <32 x i16>
+ %2 = call i16 @llvm.vector.reduce.add.v32i16(<32 x i16> %1)
+ ret i16 %2
+}
+
+define i16 @test_v64i16_v64i8(<64 x i8> %a0) {
+; SSE2-LABEL: test_v64i16_v64i8:
+; SSE2: # %bb.0:
+; SSE2-NEXT: pxor %xmm8, %xmm8
+; SSE2-NEXT: movdqa %xmm2, %xmm5
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm5 = xmm5[8],xmm8[8],xmm5[9],xmm8[9],xmm5[10],xmm8[10],xmm5[11],xmm8[11],xmm5[12],xmm8[12],xmm5[13],xmm8[13],xmm5[14],xmm8[14],xmm5[15],xmm8[15]
+; SSE2-NEXT: movdqa %xmm0, %xmm6
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm6 = xmm6[8],xmm8[8],xmm6[9],xmm8[9],xmm6[10],xmm8[10],xmm6[11],xmm8[11],xmm6[12],xmm8[12],xmm6[13],xmm8[13],xmm6[14],xmm8[14],xmm6[15],xmm8[15]
+; SSE2-NEXT: movdqa %xmm3, %xmm7
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm7 = xmm7[8],xmm8[8],xmm7[9],xmm8[9],xmm7[10],xmm8[10],xmm7[11],xmm8[11],xmm7[12],xmm8[12],xmm7[13],xmm8[13],xmm7[14],xmm8[14],xmm7[15],xmm8[15]
+; SSE2-NEXT: movdqa %xmm1, %xmm4
+; SSE2-NEXT: punpckhbw {{.*#+}} xmm4 = xmm4[8],xmm8[8],xmm4[9],xmm8[9],xmm4[10],xmm8[10],xmm4[11],xmm8[11],xmm4[12],xmm8[12],xmm4[13],xmm8[13],xmm4[14],xmm8[14],xmm4[15],xmm8[15]
+; SSE2-NEXT: paddw %xmm7, %xmm4
+; SSE2-NEXT: paddw %xmm5, %xmm4
+; SSE2-NEXT: paddw %xmm6, %xmm4
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm8[0],xmm2[1],xmm8[1],xmm2[2],xmm8[2],xmm2[3],xmm8[3],xmm2[4],xmm8[4],xmm2[5],xmm8[5],xmm2[6],xmm8[6],xmm2[7],xmm8[7]
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm8[0],xmm0[1],xmm8[1],xmm0[2],xmm8[2],xmm0[3],xmm8[3],xmm0[4],xmm8[4],xmm0[5],xmm8[5],xmm0[6],xmm8[6],xmm0[7],xmm8[7]
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm8[0],xmm3[1],xmm8[1],xmm3[2],xmm8[2],xmm3[3],xmm8[3],xmm3[4],xmm8[4],xmm3[5],xmm8[5],xmm3[6],xmm8[6],xmm3[7],xmm8[7]
+; SSE2-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm8[0],xmm1[1],xmm8[1],xmm1[2],xmm8[2],xmm1[3],xmm8[3],xmm1[4],xmm8[4],xmm1[5],xmm8[5],xmm1[6],xmm8[6],xmm1[7],xmm8[7]
+; SSE2-NEXT: paddw %xmm3, %xmm1
+; SSE2-NEXT: paddw %xmm2, %xmm1
+; SSE2-NEXT: paddw %xmm4, %xmm1
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[2,3,2,3]
+; SSE2-NEXT: paddw %xmm1, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; SSE2-NEXT: paddw %xmm0, %xmm1
+; SSE2-NEXT: movdqa %xmm1, %xmm0
+; SSE2-NEXT: psrld $16, %xmm0
+; SSE2-NEXT: paddw %xmm1, %xmm0
+; SSE2-NEXT: movd %xmm0, %eax
+; SSE2-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE2-NEXT: retq
+;
+; SSE41-LABEL: test_v64i16_v64i8:
+; SSE41: # %bb.0:
+; SSE41-NEXT: pxor %xmm4, %xmm4
+; SSE41-NEXT: pmovzxbw {{.*#+}} xmm5 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
+; SSE41-NEXT: punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm4[8],xmm2[9],xmm4[9],xmm2[10],xmm4[10],xmm2[11],xmm4[11],xmm2[12],xmm4[12],xmm2[13],xmm4[13],xmm2[14],xmm4[14],xmm2[15],xmm4[15]
+; SSE41-NEXT: pmovzxbw {{.*#+}} xmm8 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; SSE41-NEXT: punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm4[8],xmm0[9],xmm4[9],xmm0[10],xmm4[10],xmm0[11],xmm4[11],xmm0[12],xmm4[12],xmm0[13],xmm4[13],xmm0[14],xmm4[14],xmm0[15],xmm4[15]
+; SSE41-NEXT: pmovzxbw {{.*#+}} xmm7 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero,xmm3[4],zero,xmm3[5],zero,xmm3[6],zero,xmm3[7],zero
+; SSE41-NEXT: punpckhbw {{.*#+}} xmm3 = xmm3[8],xmm4[8],xmm3[9],xmm4[9],xmm3[10],xmm4[10],xmm3[11],xmm4[11],xmm3[12],xmm4[12],xmm3[13],xmm4[13],xmm3[14],xmm4[14],xmm3[15],xmm4[15]
+; SSE41-NEXT: pmovzxbw {{.*#+}} xmm6 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
+; SSE41-NEXT: punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm4[8],xmm1[9],xmm4[9],xmm1[10],xmm4[10],xmm1[11],xmm4[11],xmm1[12],xmm4[12],xmm1[13],xmm4[13],xmm1[14],xmm4[14],xmm1[15],xmm4[15]
+; SSE41-NEXT: paddw %xmm3, %xmm1
+; SSE41-NEXT: paddw %xmm2, %xmm1
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: paddw %xmm7, %xmm6
+; SSE41-NEXT: paddw %xmm5, %xmm6
+; SSE41-NEXT: paddw %xmm1, %xmm6
+; SSE41-NEXT: paddw %xmm8, %xmm6
+; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm6[2,3,2,3]
+; SSE41-NEXT: paddw %xmm6, %xmm0
+; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; SSE41-NEXT: paddw %xmm0, %xmm1
+; SSE41-NEXT: movdqa %xmm1, %xmm0
+; SSE41-NEXT: psrld $16, %xmm0
+; SSE41-NEXT: paddw %xmm1, %xmm0
+; SSE41-NEXT: movd %xmm0, %eax
+; SSE41-NEXT: # kill: def $ax killed $ax killed $eax
+; SSE41-NEXT: retq
+;
+; AVX1-SLOW-LABEL: test_v64i16_v64i8:
+; AVX1-SLOW: # %bb.0:
+; AVX1-SLOW-NEXT: vpxor %xmm2, %xmm2, %xmm2
+; AVX1-SLOW-NEXT: vpunpckhbw {{.*#+}} xmm3 = xmm1[8],xmm2[8],xmm1[9],xmm2[9],xmm1[10],xmm2[10],xmm1[11],xmm2[11],xmm1[12],xmm2[12],xmm1[13],xmm2[13],xmm1[14],xmm2[14],xmm1[15],xmm2[15]
+; AVX1-SLOW-NEXT: vpunpckhbw {{.*#+}} xmm4 = xmm0[8],xmm2[8],xmm0[9],xmm2[9],xmm0[10],xmm2[10],xmm0[11],xmm2[11],xmm0[12],xmm2[12],xmm0[13],xmm2[13],xmm0[14],xmm2[14],xmm0[15],xmm2[15]
+; AVX1-SLOW-NEXT: vextractf128 $1, %ymm1, %xmm5
+; AVX1-SLOW-NEXT: vpunpckhbw {{.*#+}} xmm6 = xmm5[8],xmm2[8],xmm5[9],xmm2[9],xmm5[10],xmm2[10],xmm5[11],xmm2[11],xmm5[12],xmm2[12],xmm5[13],xmm2[13],xmm5[14],xmm2[14],xmm5[15],xmm2[15]
+; AVX1-SLOW-NEXT: vextractf128 $1, %ymm0, %xmm7
+; AVX1-SLOW-NEXT: vpunpckhbw {{.*#+}} xmm2 = xmm7[8],xmm2[8],xmm7[9],xmm2[9],xmm7[10],xmm2[10],xmm7[11],xmm2[11],xmm7[12],xmm2[12],xmm7[13],xmm2[13],xmm7[14],xmm2[14],xmm7[15],xmm2[15]
+; AVX1-SLOW-NEXT: vpaddw %xmm6, %xmm2, %xmm2
+; AVX1-SLOW-NEXT: vpaddw %xmm2, %xmm3, %xmm2
+; AVX1-SLOW-NEXT: vpaddw %xmm2, %xmm4, %xmm2
+; AVX1-SLOW-NEXT: vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
+; AVX1-SLOW-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX1-SLOW-NEXT: vpmovzxbw {{.*#+}} xmm3 = xmm5[0],zero,xmm5[1],zero,xmm5[2],zero,xmm5[3],zero,xmm5[4],zero,xmm5[5],zero,xmm5[6],zero,xmm5[7],zero
+; AVX1-SLOW-NEXT: vpmovzxbw {{.*#+}} xmm4 = xmm7[0],zero,xmm7[1],zero,xmm7[2],zero,xmm7[3],zero,xmm7[4],zero,xmm7[5],zero,xmm7[6],zero,xmm7[7],zero
+; AVX1-SLOW-NEXT: vpaddw %xmm3, %xmm4, %xmm3
+; AVX1-SLOW-NEXT: vpaddw %xmm3, %xmm1, %xmm1
+; AVX1-SLOW-NEXT: vpaddw %xmm2, %xmm1, %xmm1
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX1-SLOW-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-SLOW-NEXT: vmovd %xmm0, %eax
+; AVX1-SLOW-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-SLOW-NEXT: vzeroupper
+; AVX1-SLOW-NEXT: retq
+;
+; AVX1-FAST-LABEL: test_v64i16_v64i8:
+; AVX1-FAST: # %bb.0:
+; AVX1-FAST-NEXT: vpxor %xmm2, %xmm2, %xmm2
+; AVX1-FAST-NEXT: vpunpckhbw {{.*#+}} xmm3 = xmm1[8],xmm2[8],xmm1[9],xmm2[9],xmm1[10],xmm2[10],xmm1[11],xmm2[11],xmm1[12],xmm2[12],xmm1[13],xmm2[13],xmm1[14],xmm2[14],xmm1[15],xmm2[15]
+; AVX1-FAST-NEXT: vpunpckhbw {{.*#+}} xmm4 = xmm0[8],xmm2[8],xmm0[9],xmm2[9],xmm0[10],xmm2[10],xmm0[11],xmm2[11],xmm0[12],xmm2[12],xmm0[13],xmm2[13],xmm0[14],xmm2[14],xmm0[15],xmm2[15]
+; AVX1-FAST-NEXT: vextractf128 $1, %ymm1, %xmm5
+; AVX1-FAST-NEXT: vpunpckhbw {{.*#+}} xmm6 = xmm5[8],xmm2[8],xmm5[9],xmm2[9],xmm5[10],xmm2[10],xmm5[11],xmm2[11],xmm5[12],xmm2[12],xmm5[13],xmm2[13],xmm5[14],xmm2[14],xmm5[15],xmm2[15]
+; AVX1-FAST-NEXT: vextractf128 $1, %ymm0, %xmm7
+; AVX1-FAST-NEXT: vpunpckhbw {{.*#+}} xmm2 = xmm7[8],xmm2[8],xmm7[9],xmm2[9],xmm7[10],xmm2[10],xmm7[11],xmm2[11],xmm7[12],xmm2[12],xmm7[13],xmm2[13],xmm7[14],xmm2[14],xmm7[15],xmm2[15]
+; AVX1-FAST-NEXT: vpaddw %xmm6, %xmm2, %xmm2
+; AVX1-FAST-NEXT: vpaddw %xmm2, %xmm3, %xmm2
+; AVX1-FAST-NEXT: vpaddw %xmm2, %xmm4, %xmm2
+; AVX1-FAST-NEXT: vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
+; AVX1-FAST-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX1-FAST-NEXT: vpmovzxbw {{.*#+}} xmm3 = xmm5[0],zero,xmm5[1],zero,xmm5[2],zero,xmm5[3],zero,xmm5[4],zero,xmm5[5],zero,xmm5[6],zero,xmm5[7],zero
+; AVX1-FAST-NEXT: vpmovzxbw {{.*#+}} xmm4 = xmm7[0],zero,xmm7[1],zero,xmm7[2],zero,xmm7[3],zero,xmm7[4],zero,xmm7[5],zero,xmm7[6],zero,xmm7[7],zero
+; AVX1-FAST-NEXT: vpaddw %xmm3, %xmm4, %xmm3
+; AVX1-FAST-NEXT: vpaddw %xmm3, %xmm1, %xmm1
+; AVX1-FAST-NEXT: vpaddw %xmm2, %xmm1, %xmm1
+; AVX1-FAST-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX1-FAST-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX1-FAST-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vphaddw %xmm0, %xmm0, %xmm0
+; AVX1-FAST-NEXT: vmovd %xmm0, %eax
+; AVX1-FAST-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX1-FAST-NEXT: vzeroupper
+; AVX1-FAST-NEXT: retq
+;
+; AVX2-LABEL: test_v64i16_v64i8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovzxbw {{.*#+}} ymm2 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero,xmm1[8],zero,xmm1[9],zero,xmm1[10],zero,xmm1[11],zero,xmm1[12],zero,xmm1[13],zero,xmm1[14],zero,xmm1[15],zero
+; AVX2-NEXT: vpmovzxbw {{.*#+}} ymm3 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
+; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm1
+; AVX2-NEXT: vpmovzxbw {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero,xmm1[8],zero,xmm1[9],zero,xmm1[10],zero,xmm1[11],zero,xmm1[12],zero,xmm1[13],zero,xmm1[14],zero,xmm1[15],zero
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm0
+; AVX2-NEXT: vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
+; AVX2-NEXT: vpaddw %ymm1, %ymm0, %ymm0
+; AVX2-NEXT: vpaddw %ymm0, %ymm2, %ymm0
+; AVX2-NEXT: vpaddw %ymm0, %ymm3, %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX2-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_v64i16_v64i8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm1
+; AVX512-NEXT: vpmovzxbw {{.*#+}} zmm1 = ymm1[0],zero,ymm1[1],zero,ymm1[2],zero,ymm1[3],zero,ymm1[4],zero,ymm1[5],zero,ymm1[6],zero,ymm1[7],zero,ymm1[8],zero,ymm1[9],zero,ymm1[10],zero,ymm1[11],zero,ymm1[12],zero,ymm1[13],zero,ymm1[14],zero,ymm1[15],zero,ymm1[16],zero,ymm1[17],zero,ymm1[18],zero,ymm1[19],zero,ymm1[20],zero,ymm1[21],zero,ymm1[22],zero,ymm1[23],zero,ymm1[24],zero,ymm1[25],zero,ymm1[26],zero,ymm1[27],zero,ymm1[28],zero,ymm1[29],zero,ymm1[30],zero,ymm1[31],zero
+; AVX512-NEXT: vpmovzxbw {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero,ymm0[16],zero,ymm0[17],zero,ymm0[18],zero,ymm0[19],zero,ymm0[20],zero,ymm0[21],zero,ymm0[22],zero,ymm0[23],zero,ymm0[24],zero,ymm0[25],zero,ymm0[26],zero,ymm0[27],zero,ymm0[28],zero,ymm0[29],zero,ymm0[30],zero,ymm0[31],zero
+; AVX512-NEXT: vpaddw %zmm1, %zmm0, %zmm0
+; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm1
+; AVX512-NEXT: vpaddw %zmm1, %zmm0, %zmm0
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vpsrld $16, %xmm0, %xmm1
+; AVX512-NEXT: vpaddw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT: vmovd %xmm0, %eax
+; AVX512-NEXT: # kill: def $ax killed $ax killed $eax
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %1 = zext <64 x i8> %a0 to <64 x i16>
+ %2 = call i16 @llvm.vector.reduce.add.v64i16(<64 x i16> %1)
+ ret i16 %2
+
+}
+
+declare i64 @llvm.vector.reduce.add.v2i64(<2 x i64>)
+declare i64 @llvm.vector.reduce.add.v4i64(<4 x i64>)
+declare i64 @llvm.vector.reduce.add.v8i64(<8 x i64>)
+declare i64 @llvm.vector.reduce.add.v16i64(<16 x i64>)
+
+declare i32 @llvm.vector.reduce.add.v2i32(<2 x i32>)
+declare i32 @llvm.vector.reduce.add.v4i32(<4 x i32>)
+declare i32 @llvm.vector.reduce.add.v8i32(<8 x i32>)
+declare i32 @llvm.vector.reduce.add.v16i32(<16 x i32>)
+declare i32 @llvm.vector.reduce.add.v32i32(<32 x i32>)
+
+declare i16 @llvm.vector.reduce.add.v2i16(<2 x i16>)
+declare i16 @llvm.vector.reduce.add.v4i16(<4 x i16>)
+declare i16 @llvm.vector.reduce.add.v8i16(<8 x i16>)
+declare i16 @llvm.vector.reduce.add.v16i16(<16 x i16>)
+declare i16 @llvm.vector.reduce.add.v32i16(<32 x i16>)
+declare i16 @llvm.vector.reduce.add.v64i16(<64 x i16>)
+
+declare i8 @llvm.vector.reduce.add.v2i8(<2 x i8>)
+declare i8 @llvm.vector.reduce.add.v4i8(<4 x i8>)
+declare i8 @llvm.vector.reduce.add.v8i8(<8 x i8>)
+declare i8 @llvm.vector.reduce.add.v16i8(<16 x i8>)
+declare i8 @llvm.vector.reduce.add.v32i8(<32 x i8>)
+declare i8 @llvm.vector.reduce.add.v64i8(<64 x i8>)
+declare i8 @llvm.vector.reduce.add.v128i8(<128 x i8>)
More information about the llvm-commits
mailing list